INTRODUCTION — Gastrointestinal toxicity due to chemotherapeutic drugs is a common problem in cancer patients. The specific chemotherapy-related gastrointestinal complications that are reviewed here include diarrhea, constipation, and intestinal perforation. Evaluation and management of patients with acute chemotherapy-related diarrhea is discussed separately, as is diarrhea related to immunotherapy with checkpoint inhibitors, chemotherapy-induced oral toxicity (mucositis), and chemotherapy-induced nausea and vomiting. (See "Management of acute chemotherapy-related diarrhea" and "Toxicities associated with checkpoint inhibitor immunotherapy", section on 'Diarrhea/colitis' and "Oral toxicity associated with chemotherapy" and "Pathophysiology and prediction of chemotherapy-induced nausea and vomiting" and "Prevention and treatment of chemotherapy-induced nausea and vomiting in adults".)

DIARRHEA — Chemotherapy-related diarrhea (CRD) is most commonly described with fluoropyrimidines (particularly fluorouracil [FU] and capecitabine) and irinotecan. Diarrhea is the dose-limiting and major toxicity of regimens containing a fluoropyrimidine with irinotecan. However, in addition to conventional cytotoxic drugs, several molecularly targeted agents (including tyrosine kinase inhibitors [TKIs] and monoclonal antibodies directed against the epidermal growth factor receptor [EGFR]) are also associated with CRD. (See 'Specific drugs' below.)

Pathogenesis/mechanisms — CRD generally occurs through three major mechanisms: increased secretion of electrolytes caused by luminal secretagogues or reduced absorptive capacity (due to surgery or epithelial damage), called secretory diarrhea; increased intraluminal osmotic substances leading to osmotic diarrhea; or altered gastrointestinal (GI) motility. Direct ischemic mucosal damage is reported in patients treated with agents targeting the vascular endothelial growth factor (VEGF), while an immune-mediated colitis is thought responsible for diarrhea with immune checkpoint inhibitors.

Secretory diarrhea — Both FU and irinotecan cause acute damage to the intestinal mucosa, leading to loss of epithelium [1,2]. FU induces mitotic arrest of crypt cells, leading to an increase in the ratio of immature secretory crypt cells to mature villous enterocytes [1,3]. The increased volume of fluid that leaves the small bowel exceeds the absorptive capacity of the colon, leading to clinically significant diarrhea. Irinotecan produces mucosal changes associated with apoptosis, such as epithelial vacuolization and goblet cell hyperplasia, suggestive of mucin hypersecretion [2]. These changes appear related to the accumulation of the active metabolite of irinotecan, SN-38, in the intestinal mucosa [4].

Up to 50 percent of patients treated with TKIs experience diarrhea [5]. It is thought that the diarrhea occurs through multiple mechanisms. Increased chloride secretion caused by dysregulation of the EGFR signaling pathway, [6,7] colonic crypt damage, gut dysmotility, and alteration in gut microbiota have been proposed. Monoclonal antibodies that release cancer-induced suppression of the immune system, such as ipilimumab, a human antibody to cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), cause autoimmune colitis. Diffuse, segmental, or patchy colitis is seen on colonoscopy. Histologically, nonspecific acute and chronic inflammatory infiltrate, cryptitis, and crypt abscesses are noted [8]. Colonic perforation (less than 1 percent) and death in 5 percent of patients have been reported. Rituximab, an anti-CD20 monoclonal antibody used to treat B-cell lymphoma, can cause new-onset ulcerative colitis or exacerbation of preexisting colitis [9].

Osmotic diarrhea — Damage to the brush border enzyme system within the epithelium can cause diarrhea. Carbohydrates and proteins need to be digested before they are absorbed through the villi, and fats require emulsification. Approximately 10 percent of patients being treated with FU have decreased expression of the enzyme lactase in their intestinal brush border, leading to lactose intolerance [10,11]. The D-xylose absorption test has been reported to be abnormal in several patients undergoing chemotherapy [12,13], suggesting the presence of carbohydrate malabsorption (such as sucrose, fructose, or even complex polysaccharides).

Altered intestinal motility — Early-onset diarrhea with irinotecan occurs during or within several hours of drug infusion in 45 to 50 percent of patients and is cholinergically mediated [14]. In contrast, late irinotecan-associated diarrhea is not cholinergically mediated and, instead, appears to be multifactorial, with contributions from dysmotility and secretory factors, as well as a direct toxic effect on the intestinal mucosa. (See 'Irinotecan' below.)

Antiangiogenesis therapy — Antiangiogenesis therapy is associated with GI tract perforation [15-17]. The mechanism by which this happens has not been proven, but proposed mechanisms include intestinal wall disruption (ulceration) in areas of tumor necrosis, disturbed platelet-endothelial cell homeostasis causing submucosal inflammation and subsequent ulcer formation, impaired healing of pathologic, chemotherapy-induced or surgical bowel injury, and mesenteric ischemia from thrombosis and/or vasoconstriction; all of these may result in diarrhea as well. Randomized trials of bevacizumab added to chemotherapy for colorectal cancer have resulted in small increases in severe diarrhea, but this has not been observed in other tumor types, suggesting that bevacizumab may not be causing diarrhea directly but, instead, by one of the indirect mechanisms mentioned above [18,19]. (See "Toxicity of molecularly targeted antiangiogenic agents: Non-cardiovascular effects", section on 'Intestinal perforation/fistula formation'.)

Checkpoint inhibitor immunotherapy — Ipilimumab is a monoclonal antibody directed against CTLA-4, a molecule on the T-cell surface. Pembrolizumab and nivolumab are monoclonal antibodies directed against the programmed cell death 1 protein (PD-1). All three drugs are approved for immunotherapy of advanced melanoma. Nivolumab is also approved for advanced non-small cell lung cancer, renal cell carcinoma, and Hodgkin lymphoma. Pembrolizumab is also approved for non-small cell lung cancer and head and neck squamous cell cancer. (See "Immunotherapy of advanced melanoma with immune checkpoint inhibition", section on 'Ipilimumab' and "Immunotherapy of non-small cell lung cancer with immune checkpoint inhibition", section on 'Nivolumab' and "Immunotherapy of advanced melanoma with immune checkpoint inhibition".)

The presumed mechanism underlying checkpoint inhibitor immunotherapy is to break down tolerance to the tumor-associated antigens in the melanoma. At the same time, this may result in decreased tolerance to self-antigens. A wide range of immune-mediated adverse events have been observed in patients treated with both drugs, including enterocolitis, which can be serious or life-threatening. It is critical to recognize that treatment of diarrhea caused by ipilimumab, nivolumab, or pembrolizumab may require systemic corticosteroids. The manifestations and management of ipilimumab-, nivolumab-, and pembrolizumab-induced enterocolitis are discussed separately. (See "Toxicities associated with checkpoint inhibitor immunotherapy", section on 'Diarrhea/colitis'.)

Clinical manifestations — Typical CRD typically begins with an increasing frequency of bowel movements and/or a loosening of stool consistency. Excessive gas and/or intestinal cramping commonly accompanies CRD. As the CRD progresses, it can become severe, with frequent watery stools. CRD can be debilitating and, in some cases, life-threatening. Findings in such patients include volume depletion, acute kidney injury, and electrolyte disorders, such as hypokalemia, metabolic acidosis, and depending upon water intake, hyponatremia (increased water intake that cannot be excreted because of the hypovolemic stimulus to release of antidiuretic hormone) or hypernatremia (insufficient water intake to replace losses). Infection, including life-threatening sepsis, can result due to a breach of the intestinal mucosa, which is worsened in the setting of chemotherapy-induced immunosuppression.

Given the risk for dehydration and infection, severe CRD frequently requires hospital admission for adequate supportive care [20-23]. Other sequelae of CRD include increased cost of care, reduced quality of life, treatment delays, and diminished compliance with treatment regimens, which may compromise long-term outcomes if the chemotherapy is being administered with curative intent [21].

The severity of CRD is often described using the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) grades; the latest version is outlined in the table (table 1). Severity is determined by the number of stools per day or an increase in ostomy output compared to baseline, the need for hospitalization, and the effect on activities of self-care. It is critical to ascertain the patient's baseline bowel pattern when grading the severity of diarrhea.

Assessment of patients with acute CRD is addressed in detail elsewhere. (See "Management of acute chemotherapy-related diarrhea", section on 'Clinical assessment'.)

Colitis syndrome that may arise in patients treated with chemotherapy — Patients undergoing chemotherapy may develop several types of colitis.

●Neutropenic enterocolitis – Neutropenic enterocolitis (a form of necrotizing enterocolitis or typhlitis) is one of the most common GI complications in leukemic patients who are undergoing induction therapy, and can occur in other malignancies and following stem-cell-supported high-dose chemotherapy [24,25].

Neutropenic enterocolitis should be considered in any severely neutropenic patient (absolute neutrophil count <500 cells/microL) who presents with fever and abdominal pain. The location of abdominal pain is often in the right lower quadrant, and symptoms, including fever, frequently appear during the third week (median 17 days) after receiving cytotoxic chemotherapy, at a time when neutropenia is most profound. Additional symptoms may include abdominal distension, cramping, tenderness, nausea, vomiting, watery or bloody diarrhea, and frank hematochezia.

The pathogenesis, risk factors, diagnosis, and management of patients with neutropenic enterocolitis are discussed separately. (See "Neutropenic enterocolitis (typhlitis)".)

●Ischemic colitis – A small number of cases of ischemic colitis have been reported with docetaxel-containing regimens in patients with metastatic breast cancer, including 3 of 14 patients in a phase I study of docetaxel plus vinorelbine [26,27]. The typical onset is 4 to 10 days following administration. Patients with acute colonic ischemia usually present with rapid onset of mild abdominal pain and tenderness over the affected bowel, commonly on the left side of the abdomen. Mild to moderate amounts of rectal bleeding or bloody diarrhea typically develop within 24 hours of the onset of abdominal pain. (See "Overview of intestinal ischemia in adults".)

●Clostridium difficile-associated colitis – Clostridium difficile (C. difficile) colitis is a common problem in patients with cancer, mostly due to the high rate of oral antibiotic use and hospitalization. However, several reports have described this complication in patients without any prior antibiotic use following chemotherapy [28-30]. The proposed mechanism is chemotherapy-induced intestinal damage that facilitates the proliferation of C. difficile. Frequent occurrence of C. difficile-related diarrheal episodes has been reported in patients treated with paclitaxel-containing regimens, especially with the use of "dose-dense" regimens [31].

Watery diarrhea is the cardinal symptom of C. difficile–associated diarrhea with colitis (≥3 loose stools in 24 hours). Other manifestations include lower abdominal pain and cramping, low-grade fever, nausea, anorexia, and leukocytosis. The spectrum of illness varies from mild to fulminant; management is discussed in detail elsewhere. (See "Clostridium difficile infection in adults: Clinical manifestations and diagnosis" and "Clostridium difficile infection in adults: Treatment and prevention".)

Differential diagnosis — Patients who develop acute diarrhea during chemotherapy may also suffer from organic causes of diarrhea, such as bacterial overgrowth, fat or bile acid malabsorption, intake of excess quantities of sorbitol or lactose intolerance, and inflammatory and infectious causes, which should not be overlooked (see "Management of acute chemotherapy-related diarrhea", section on 'Differential diagnosis'). These include:

●Small intestinal bacterial overgrowth

●Fat malabsorption due to causes other than chemotherapy

●Bile acid malabsorption or bile acid diarrhea

●Other causes of osmotic diarrhea (lactose intolerance, laxative overuse)

●Infectious causes, especially C. difficile

●Stool impaction or tumor-related obstruction

Specific drugs

Fluorouracil — One of the main dose-limiting side effects of fluoropyrimidines such as FU (and its oral derivatives capecitabine and ftorafur-uracil [UFT], see below) is diarrhea. Both the therapeutic efficacy and frequency of diarrhea associated with FU are increased when given concurrently with leucovorin (LV). Diarrhea is also schedule-dependent. The highest frequency of diarrhea occurs with bolus rather than continuous 24-hour infusion of FU/LV for up to five days, particularly with weekly bolus administration, but it can occur with all schedules of administration [32-37]. (See "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'Leucovorin plus FU' and "Adjuvant therapy for resected stage III (node-positive) colon cancer", section on 'Fluoropyrimidine-based therapy'.)

In multiple reports of weekly FU/LV, diarrhea has been seen in up to 50 percent of patients, one-half of whom required hospitalization for intravenous (IV) fluids [32,33,36]. In one series, 11 of 221 patients (5 percent) died of treatment-related toxicity, most of them older adult patients with concomitant leukopenia and sepsis [36]. Based upon these observations, treatment is routinely withheld for grade 2 or worse (table 1) diarrhea and not restarted until diarrhea resolves, an approach that has led to a significant reduction in severe enterotoxicity with these regimens. (See "Management of acute chemotherapy-related diarrhea", section on 'Restarting chemotherapy'.)

No risk factor can reliably predict the development of diarrhea with FU therapy, but the risk is increased in the presence of an unresected primary tumor, shortened bowel resulting from previous surgery, previous episodes of CRD, treatment during the summer season [38], and when bolus FU and LV are combined with oxaliplatin [39] or irinotecan. (See 'IFL' below.)

Women appear to suffer more toxicity from FU than men for unclear reasons [40,41]. In a meta-analysis of over 2400 patients enrolled in five trials (three for advanced disease, two in the adjuvant setting), significantly more women experienced ≥grade 3 toxicity during therapy than did men (50 versus 40 percent) [40]. All of these trials used a five-day bolus schedule of FU and LV. These data suggest that women may be better served by less toxic administration schedules of FU, including short-term infusional regimens. (See "Adjuvant therapy for resected stage III (node-positive) colon cancer" and "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'Leucovorin plus FU'.)

Predictive markers — Early detection of patients who are at risk of developing life-threatening toxicity from fluoropyrimidines based upon predictive markers might allow dose reduction or selection of an alternative treatment regimen. The two most well-studied predictive factors are enzymatic activity of dihydropyrimidine dehydrogenase (DPD) and thymidylate synthetase (table 2).

DPD deficiency — Dihydropyrimidine dehydrogenase (DPD), the first of three enzymes in the fluoropyrimidine metabolic pathway, is the rate-limiting enzyme in FU catabolism. Enzyme activity varies widely, with most of the variability arising from genetic polymorphisms in the dihydropyrimidine dehydrogenase (DPYD) gene. Patients who are partially or totally deficient in DPD activity cannot adequately degrade fluoropyrimidines, leading to an increased risk of severe, sometimes fatal, toxicity. In patients with even partial DPD deficiency, administration of a fluoropyrimidine can lead to life-threatening complications, including severe diarrhea, mucositis, and pancytopenia [42-50]. Nausea, vomiting, rectal bleeding, volume depletion, skin changes, neurologic abnormalities (cerebellar ataxia, cognitive dysfunction, altered level of consciousness), and cardiotoxicity may also occur. (See "Fluoropyrimidine-associated cardiotoxicity: Incidence, clinical manifestations, mechanisms, and management", section on 'DPYD polymorphisms'.)

Although complete DPD deficiency is rare (and usually associated with homozygosity for one of the alleles associated with reduced enzyme activity), reduced levels of enzyme activity due to the inheritance of one high-risk allele are more common, particularly in black women. These demographic differences can be illustrated by an analysis of enzyme levels from 258 normal volunteers [51]. No patient had complete DPD deficiency, while partial DPD deficiency was present in 12.3, 4.0, 3.5, and 1.9 percent of black women and men, and white women and men, respectively.

Molecular analysis of patients with DPD deficiency has identified over 128 mutations and polymorphisms in the DPD gene (DPYD) that may result in partial or total loss of DPD activity [52-54]. Only four have been consistently associated with a marked decrease in DPD activity and enhanced fluoropyrimidine toxicity:

●DPYD*2A single-nucleotide polymorphism (SNP; splice site variant IVS14+1G>A with the nucleotide change 1905+1G>A, resulting in an exon 14 deletion [del]) [52,55-64].

●DPYD*13 SNP, with the nucleotide change 1679T>G, with resultant amino acid substitution I560S [52,54,65].

●DPYD*9B SNP, with the nucleotide change c.2846A>T, with resultant amino acid substitution D949V [52,54,64].

●A collection of SNPs, termed HapB3, that is composed of three intronic variants (c.483+18GG>A, c.680+139G>A, and c.959-51T>C) and one synonymous variant (c.1236G>A) has been suggested to also contribute to fluoropyrimidine toxicity. Subsequently, an additional variant in linkage disequilibrium with HapB3 (the intronic polymorphism c.1129-5923C>G) was suggested to activate a splice donor site within intron 10 of the DPYD gene and promote alternative splicing to alter the function of the gene [66-68]. However, the effects of this variant are modest [69], which probably explains the disparate reports on its influence on FU toxicity [65-68,70]. Nevertheless, in Europeans, HapB3 with c.11129-5923C>G is the most common decreased function DPYD variant, with carrier frequencies of approximately 5 percent [71].

However, at least in the United States, one of the three high-risk alleles, DPYD*2A, *13, and *9B, is present in fewer than 10 percent of patients in most populations, and inheritance of a high-risk allele is not always associated with life-threatening toxicity (ie, the positive predictive value of having one of these alleles on the risk of severe toxicity is variable). Inheritance of one of these high-risk alleles does not account for the majority of FU-associated severe toxicity, which is estimated to occur in 15 to 30 percent of treated patients (ie, sensitivity is limited). Finally, even if preemptive pharmacogenomics testing is done and these high-risk alleles are not detected, patients can still have life-threatening toxicity (ie, the specificity is limited). These issues can be illustrated by the following data:

●In one series, one of the three high-risk variants (DPYD*2A, *13, or *9B) was found in only 30 percent of FU-treated patients (13 of 44) who developed grade 3 or 4 toxicity after treatment initiation [52]. Similarly, a systematic review of the literature by the Clinical Pharmacogenetics Implementation Consortium (CPIC) concluded that between 23 and 38 percent of severe fluoropyrimidine toxicity could be attributed to DPYD variants (clinical sensitivity approximately 31 percent) [72].

●In a prospective study of 683 patients receiving FU monotherapy, grade 3 or 4 toxicity occurred in 16 percent, and genotyping revealed the DPYD*2A allele in only 5 percent of those with treatment-related toxicity [57]. Furthermore, fewer than half of those who had the DPYD*2A allele developed grade 3 or 4 toxicity (positive predictive value 46 percent). Of interest, there was a gene-sex interaction, resulting in an odds ratio [OR] of 41.8 for male patients but only 1.33 for female patients.

●In a large series of 430 patients initiating therapy with FU or capecitabine for any tumor type, 24 of the 104 patients experiencing grade 3 or 4 toxicity in the first four cycles of therapy (23 percent) were found to have one of four high-risk variant alleles (DYPD*2A, *13, *9B, and one other high-risk allele, 1601G>A) [62]. However, only 6 percent of the entire cohort had one of these four rare variants. However, in contrast to the prior series, the positive predictive value of having inherited any of these four high-risk alleles for severe (grade 3 or 4) diarrhea, mucositis, or myelosuppression during the first four cycles of therapy in this study was >99 percent.

Some of the inconsistency in study results may be attributable to variations in the treatment regimens across studies. The DPYD*2A allele has been associated with toxicity more frequently when FU was administered in combination with other chemotherapeutic agents rather than as monotherapy [72].

A major issue is that inheritance of one of these high-risk alleles does not account for all cases of DPD deficiency. Impaired DPD activity has been detected in some patients with normal wild-type DPYD alleles, presumably due to epigenetic mechanisms, including microRNAs, that regulate enzyme activity [73,74].

Despite these difficulties, the potential benefits of identifying patients who have inherited one of these three high-risk alleles in terms of reducing treatment-related toxicity and improving the cost-effectiveness of care can be illustrated by the following reports:

●In a report of data from a large cohort of patients with stage III colon cancer who were enrolled on the adjuvant NCCTG N0147 trial, the frequency of finding one of the three high-risk DPYD alleles was low overall; of the 2886 patients who were genotyped, 27 (0.9 percent), 4 (0.1 percent), and 32 (1.1 percent) carried the DPYD*2A, *13, and *9B variants, respectively [75]. However, of the 2594 patients with complete adverse event data, grade 3 or 4 adverse events developed in 22 of the 25 patients with the *2A variant (88 percent positive predictive value), in two of four *13 carriers (50 percent), and in 22 of 27 *9B carriers (82 percent). The *2A variant was significantly associated with nausea and vomiting and neutropenia but not diarrhea, while the *9B variant was significantly associated with dehydration, diarrhea, neutropenia, and thrombocytopenia.

●In another series, 2038 patients initiating fluoropyrimidine therapy were prospectively screened for DPYD*2A prior to initiating therapy; 18 of 22 patients (1.1 percent of the total) were identified as having the *2A allele and were treated with a reduced dose intensity of capecitabine (median 48 percent of usual, range 17 to 91 percent) [63]. The risk of grade 3 or worse toxicity was 28 percent in this group and compared favorably with the risk of grade 3 or 4 toxicity in a historical cohort of patients with the DYPD*2A variant who received full-dose therapy (78 percent), and the risk of drug-induced death was reduced from 10 to 0 percent. That dose reduction had not significantly altered the therapeutic efficacy of capecitabine was suggested by the comparable rate of grade 3 or worse treatment-related toxicity in DPYD wild-type patients who received standard-dose therapy (23 percent). Furthermore, the average total treatment cost per patient was modestly lower for screening than for non-screening, outweighing the cost of screening.

●In a cost-effectiveness analysis of screening patients intended to receive fluoropyrimidine-based chemotherapy just for the DPYD*2A allele, which was based on an analysis of the prevalence of DPYD*2A mutations (roughly 1 percent) and the likelihood of grade 5 toxicity (death) in this same cohort [61], the number needed to screen to avoid one FU-related death was approximately 1000. If the cost for DPYD*2A genotyping in the United States is approximately $82 per patient, it would cost $82,000 to prevent one iatrogenic death through DPYD*2A screening [76], a value that compares favorably with other medical procedures and tests. (See "A short primer on cost-effectiveness analysis".)

As noted above, the DPYD*2A, DPYD*13, DPYD*9B, and HapB3 variants are the most commonly reported high-risk DPYD variants. Other alleles have been identified that might increase the likelihood of fluoropyrimidine-related toxicity [54,62,65,77]. The data are inconsistent for one of these, the DPYD*5 allele (nucleotide change 496A>G, resulting in the M166V amino acid substitution), with some studies suggesting reduced activity and others increased enzyme activity relative to wild-type DPD [54,78-80]. Similarly, the impact of polymorphisms in the DPYD*4 allele (nucleotide change 1601G>A, resulting in amino acid substitution S534N) is also unclear.

To avoid the risk of severe and potentially fatal reactions, the manufacturer of the oral fluoropyrimidine capecitabine recommends to withhold or permanently discontinue the drug in patients with evidence of acute early-onset or unusually severe toxicity, which may indicate a near-complete or total absence of DPD activity. Furthermore, they state that no capecitabine dose has been proven safe for patients with complete absence of DPD activity and that there is insufficient data to recommend a specific dose in patients with partial DPD activity as measured by any specific test. However, they stop short of recommending preemptive testing for all patients prior to initiating therapy. (See 'Pharmacogenetic testing for DPYD and TYMS variants' below.)

If a high-risk DPYD variant is identified prior to treatment, guidelines for fluoropyrimidine dosing are available from the CPIC (table 3) [81]. However, they do not provide recommendations on whether and when to perform pharmacogenomics testing. Given the low frequency of finding a predictive allele and the low sensitivity (ie, patients who lack one of these high-risk DPYD variants may still suffer grade 3 or 4 fluoropyrimidine-related toxicity), preemptive genetic testing of all patients due to receive a fluoropyrimidine in order to identify those with DPD deficiency is controversial and not widely practiced [82]. Neither the US Food and Drug Administration (FDA) nor the European Medicines Agency (EMA) currently require pharmacogenetic testing before fluoropyrimidine administration. Testing is usually reserved for patients who develop unusually early, severe, treatment-related toxicity (diarrhea, mucositis, myelosuppression, neurotoxicity, cardiotoxicity) in the first few cycles of fluoropyrimidine therapy. This subject is discussed in more detail below. (See 'Pharmacogenetic testing for DPYD and TYMS variants' below.)

TYMS gene variations — In addition to DPYD, The available data suggest that high-risk polymorphisms in the thymidylate synthetase gene (TYMS) may be associated with a 1.4 to 2.4-fold increase in the risk of severe toxicity from FU-based chemotherapy; however, the data are less certain than with the high-risk DPYD genotypes.

Thymidylate synthetase (TS), a critical enzyme for thymidine production, is potently inhibited by FU. The FU metabolite, 5-FdUMP, forms a stable complex with TS and folate, blocking the activity of this enzyme. The important polymorphisms that might influence fluoropyrimidine toxicity are outlined in the table (table 2) and are described in detail in the following sections:

●Expression of TYMS is regulated by transcription factors that bind to the promoter region, a section of coding sequence that lies upstream from the gene. The 5’ untranslated region (UTR) contains a variable number of 28 bp tandem repeats (VNTRs) which act to enhance the promoter and stimulate transcriptional activity. The vast majority of individuals carry TYMS alleles that contain two or three repeats in this promoter region, designated 2R and 3R. These polymorphisms affect TS levels [83,84]. Patients who are homozygous for the triple repeat (3R/3R) have a greater number of binding sites for transcription factor and higher TS levels compared with those who are 2R/2R or 3R/2R; conversely, 2R/2R homozygotes have low TS levels in normal tissues and may be at a greater risk of FU cytotoxicity. Notably, patients who overexpress TS have relative resistance to fluoropyrimidines.

●An SNP has been described in the 12^th nucleotide of the second repeat of the 3R allele, which abolishes a binding site in the 3R second repeat allele (designated the 3RC allele) [83,85]. This markedly reduces TS activity to a level lower than seen with the wild-type 3R allele (designated 3RC to distinguish it from the wild-type allele 3RG) [86-88]. Others describe greater toxicity with a similar substitution of cytosine for guanine at the 12^th nucleotide in two 28 bp repeats in the 5’ UTR (designated 2RC) [87,89]

●Furthermore, in addition to these polymorphisms in the promoter region, a genetic polymorphism in the 3’ UTR of the TYMS gene has also been identified, usually presented as a del of 6 bp at position 1494 [90], which also results in low TS expression and may be associated with a greater risk of fluoropyrimidine toxicity [62].

The available data linking these polymorphisms to increased fluoropyrimidine toxicity are conflicting, as evidenced by the following reports:

●The 2R/2R genotype has been associated with greater toxicity in many [84,91-93] (but not all [62,93-95]) studies. Even in positive studies, the sensitivity and positive predictive value appear to be limited. As an example, in three studies totaling 200 unselected patients who received FU, 44 (22 percent) developed grade 3 or 4 toxicity [84,91,92]. Only 13 of the 44 had the high-risk 2R/2R genotype (sensitivity 30 percent), while of the 25 patients who inherited the 2R/2R high-risk genotype, only 13 developed grade 3 or 4 toxicity (positive predictive value 52 percent).

Not all studies consistently show greater toxicity with inheritance of 2R alleles. One prospective study attempted to use TYMS genotyping to select the chemotherapy approach to neoadjuvant chemoradiotherapy in patients with rectal cancer [95]. As noted above, higher levels of TS expression (as seen with the 3R allele versus two 2R alleles) have been associated with resistance to FU and less toxicity. In this trial, 135 patients were assigned therapy based upon their "risk" for a good response to neoadjuvant chemoradiotherapy. Patients with "good-risk" tumors (2R/2R, 2R/3R, or 2R/4R) were treated with standard fluoropyrimidine chemoradiotherapy, while those who were judged "poor risk" (3R/3R or 3R/4R) received FU in addition to irinotecan concomitant with RT. Despite inheriting a genotype that predicted for a poorer response to FU, complete pathologic responses were twice as high in the poor-risk group (42 versus 20 percent), suggesting that the addition of irinotecan overcame relative FU resistance. However, notably, rates of grade 3 and 4 toxicity were not worse in those in the good-risk group who were heterozygous or homozygous for 2R (30 versus 54 percent in the poor-risk group), and hospitalization rates were actually lower (16 versus 34 percent).

●Inheritance of a high-risk 2RC allele has been associated with an increase in toxicity [87,89]. In a series of 1613 patients, 28 had the 2RC variant allele (1.7 percent), and 20 had a high-risk genotype (2RG/2RC, 3RC/2RC, and 2RC/2RC); patients with the 3RG/2RC genotype were excluded because of higher TS levels with the 3RG genotype [89]. Both early severe toxicity and toxicity-related hospitalization were more frequent in risk-associated genotype carriers (OR 3.0 [95% CI 1.04-8.93] and 3.8 [95% CI 1.19-11.9], respectively).

●The contribution of the 3’ UTR del 6 bp variant is uncertain. In one study, the presence of a homozygous del/del genotype approximately doubled the risk of grade 3 or 4 toxicity [62]. The sensitivity was 47 percent, the specificity was 54 percent, and the positive predictive value was only 25 percent. Others have failed to confirm greater toxicity with this variant [94,96,97].

Other predictive markers — Deficiencies have also been reported in other catabolic enzymes (dihydropyrimidinase and beta-ureidopropionase) that impair the uracil catabolic pathway [98,99]. Their contribution to FU toxicity is not well characterized.

The contribution of other genetic and nongenetic factors has not been well studied [96,100]. At least some data suggest marked regional differences in tolerability of fluoropyrimidines, with the highest toxicity rates in the United States and the lowest in East Asia [101]. This may be due, in part, to differences in dietary folic acid intake [101,102]. Reduced folates (such as LV) stabilize the binding of the FU metabolite fluorodeoxyuridine monophosphate to thymidylate synthetase, enhancing the response to fluoropyrimidine therapy [103] but also increasing toxicity. Higher pretreatment serum folate levels have also been linked to greater toxicity from capecitabine, an orally active prodrug of FU [104]. (See 'Capecitabine' below.)

A model combining genetic and nongenetic factors has been developed that predicted severe enterotoxicity in patients who were mostly treated with FU monotherapy [57]. However, the model has been criticized for its lack of comprehensiveness, the diverse group of patients studied, and the use of FU monotherapy, which is uncommonly used [100].

Pharmacogenetic testing for DPYD and TYMS variants — Pharmacogenetic profiling has the potential to identify patients who may experience severe adverse effects with fluoropyrimidines. However, preemptive testing has not been widely adopted in routine clinical care, and it is not recommended as a routine test in treatment guidelines for metastatic colorectal cancer from the European Society of Medical Oncology (ESMO) [105] prior to the initiation of a fluoropyrimidine. Although genotyping may identify a small fraction of patients for whom serious toxicity is a concern because of inherited variants in DPYD or TYMS, approximately one-half of patients who develop serious toxicity to fluoropyrimidines do not have an identifiable mutation or polymorphism in either gene (ie, sensitivity is limited). Furthermore, the frequency of these alleles in the general population is quite small, and not all patients who inherit these genetic mutations or polymorphisms suffer from life-threatening toxicity (ie, the positive predictive value is variable). For these reasons, at many institutions, genotyping is reserved for those patients who have unexpected toxicity (myelosuppression, mucositis, diarrhea, neurotoxicity, cardiotoxicity) during the first few cycles of fluoropyrimidine therapy.

Diagnosis of DPD deficiency can be made by radioimmunometric assay for the DPD enzyme [106] or, more commonly, by sequencing of the DPYD gene from a sample of peripheral blood. Current guidelines for DPYD testing from CPIC include only specific variants in DPYD (DPYD*2A, *13, and *9B), which are thought to be the most important but probably account for only some of the variants that are associated with a decreased ability to metabolize these drugs [72].

Neither test provides rapid results (average turnaround time is 13.3 days) [107]. A simple uracil breath test (2-13C uracil breath test [UraBT]) has been developed to detect abnormalities in the entire DPD catabolic pathway, and rapid plasma tests for the ratio of uracil to 5,6-dihydrouracil:uracil (U:UH2) or serum uracil concentration surrogates for DPD activity have also been developed to discriminate individuals who manifest severe toxicity from FU [50,108-111], but neither is commercially available. It is also unclear whether either the ratio of U:UH2 or uracil concentrations alone correlate with systemic DPD activity and risk of fluoropyrimidine-related toxicity [112].

The original test (TheraGuide FU, Myriad Laboratories) that analyzed DNA from peripheral blood cells to fully sequence the DPYD gene (assessing the presence of the three most common DYPD polymorphisms and other rare variants) and detected TYMS variants that increase risk for FU toxicity has been discontinued. In its place, a new assay has been developed (FU Toxicity and Chemotherapeutic Response Panel, 5 mutations, ARUP Laboratories). It uses polymerase chain reaction (PCR)/fluorescence monitoring to detect the three common high-risk DPYD variants (*2A [c.1905 +1 G>A], *13 [c.1679T>G, also known as I560S], and *9B [c.2846A>T, also known as D949V]) and uses PCR and restriction digest followed by capillary electrophoresis to detect a germline variant in the TYMS 3’ UTR (a 6 bp del) and tandem repeats in the 5’-promoter enhancer region (5’-TSER; the 2R, 3RG, and 3RC genotypes). Other tests are also commercially available.

Some authors suggest that genotype testing should ideally be carried out before initiation of treatment to identify those patients at high risk of toxicity with FU or capecitabine, and some institutions have adopted this approach [107,113,114]. However, this is a controversial area, and this preemptive pharmacogenetic testing has not been widely adopted. Although there are some uncontrolled data suggesting lower toxicity when FU doses are empirically adjusted based upon U:UH2 (a surrogate for DPD enzyme activity) [50] or a combined approach that included genotyping for DPYD SNPs and U:UH2 ratios [115], there are no prospective randomized trials that demonstrate improved toxicity without compromise of efficacy in patients who are preemptively screened versus not screened using any assay for DPD activity, including pharmacogenetic testing for high-risk DPYD and TYMS variants. Furthermore, while suggested recommendations for managing fluoropyrimidine drugs based upon inheritance of polymorphisms DPYD*2A, *13, and *9B are available from CPIC (table 3) [72], there are no accepted guidelines for management of patients who are identified as having a high or moderate risk for toxicity according to TYMS variants (ie, should they receive a lower initial dose, enhanced surveillance, pharmacokinetic monitoring, an alternate agent, or a more informed discussion regarding risk?). Any patient who experiences severe toxicity following FU treatment will require a significant dose reduction if treatment will continue. Patients who have unusually severe toxicity or severe toxicity within days after dosing with a fluoropyrimidine should be suspected of having an at-risk DPYD or TYMS mutation. Testing for at-risk mutations in DPYD and TYMS is reasonable in these patients, as dosing recommendations for DYPD mutation carriers are available from CPIC, and because identification of at-risk mutations may be of use if family members are to be treated in the future with fluoropyrimidines.

Management of DPD-deficient patients — If a DPYD variant is identified prior to treatment, guidelines for management are available from CPIC (table 3) [66]. The authors note that not all patients who harbor reduced- or no-function variants of DPYD manifest toxicity, and therefore, they recommend increasing the fluoropyrimidine dose in the absence of toxicity or in patients who have subtherapeutic plasma levels. This dose increase is of particular importance for patients being treated with curative intent.

Alternative agents are needed for patients who are homozygous (ie, carrying two nonfunctional alleles). The quinazoline folate analog raltitrexed, which is a thymidylate synthetase inhibitor, may be a useful substitute for FU in patients with DPD deficiency [116], but it is not available in the United States. UFT is not a safe substitute for FU in this situation [117] as it is a combination of ftorafur (tegafur), a FU pro-drug, plus uracil, which competes with FU for DPD.

Where available, another option is close monitoring of FU levels and pharmacokinetically guided dosing. (See "Dosing of anticancer agents in adults", section on 'Therapeutic drug monitoring'.)

Guidelines are not available from the CPIC or any other group for management of patients who are identified as having high-risk TYMS variants.

Most cases of DPD deficiency are diagnosed only after a severe reaction to FU. Management of these patients should include aggressive hemodynamic support, parenteral nutrition, antibiotics, hematopoietic colony stimulating factors, and where available, uridine triacetate (see 'Uridine triacetate' below). Dialysis is of no benefit if renal function is normal, since even with complete DPYD deficiency, FU is rapidly cleared through the urine [42].

Uridine triacetate — Uridine triacetate (originally called vistonuridine) is an orally administered prodrug of uridine, a specific pharmacologic antidote to fluoropyrimidines, including FU and capecitabine. It is a safe and potentially life-saving treatment for overdoses of these agents. Uridine triacetate was studied in 173 adult and pediatric patients who were treated in two separate trials and had either received an overdose of FU or capecitabine (n = 147) or had early-onset, unusually severe, or rapid-onset life-threatening toxicities within 96 hours after receiving FU (n = 26, the fraction who had DPD deficiency as the cause for severe early toxicity could not be determined) [118]. Overall, 137 of 142 assessable overdose patients treated with uridine triacetate (96 percent) survived to 30 days, had rapid reversal of acute neurotoxicity or cardiotoxicity (affecting 12 patients), and either prevention of or recovery from severe mucositis or leucopenia. Among the 26 patients treated for early-onset toxicity following fluoropyrimidine therapy (some of which presumably had DPD deficiency), 21 survived to 30 days (81 percent); all five deaths were in patients who initiated uridine triacetate beyond 96 hours after the last dose of the fluoropyrimidine. Adverse events attributable to uridine triacetate were mild and infrequent, and included diarrhea, nausea, and vomiting.

There are few data on the specific use of uridine triacetate in patients who develop severe fluoropyrimidine toxicity because of DPD deficiency [119]. However, the drug has been shown to prevent fatalities in mice who are treated with FU after receiving an inhibitor of DPD [120]. Thus, DPD-deficient patients who develop early severe toxicity after receiving the first dose of a fluoropyrimidine could also benefit from treatment with uridine triacetate, if the deficiency is identified soon enough after the drug is administered and the drug can be obtained within 96 hours of the last dose. Uridine triacetate should not be administered for nonemergency toxicities as it may interfere with the efficacy of fluoropyrimidine treatment.

Uridine triacetate has received orphan drug designation for treatment of FU overexposure from the EMA. Uridine triacetate was approved by the FDA in December 2015 for emergency use following an FU or capecitabine overdose, regardless of the presence of symptoms, for patients who exhibit early-onset, severe, or life-threatening toxicity affecting the cardiac or central nervous system, and/or early-onset, unusually severe adverse reactions (eg, GI toxicity and/or neutropenia) within 96 hours following the end of FU or capecitabine administration [121]. The recommended dose and schedule for adults is 10 g orally every six hours for 20 doses. The recommended dose and schedule for pediatric patients is 6.2 g/m² of body surface area orally every six hours for 20 doses. Despite its approval, uridine triacetate is not available commercially in any country. Ordering information for emergency use of uridine triacetate is available from vistogard.com.

Capecitabine — Capecitabine is a rationally designed oral fluoropyrimidine that is converted to FU in three sequential enzymatic reactions. The dose-limiting toxicities are diarrhea, palmar-plantar erythrodysesthesia, and neutropenia. Like FU, capecitabine is catabolized by DPD, and there is a risk for early and severe toxicity in those who are DPD deficient. However, the specific genetic markers of capecitabine-related toxicity are less well studied than with other fluoropyrimidines, such as FU. One analysis concluded that a panel of genetic biomarkers for toxicity from capecitabine monotherapy would include only two DPYD and two TYMS variants; when only these polymorphisms were assessed, the sensitivity for identifying those with severe toxicity was only 26 percent, and the positive predictive value was only 49 percent; the negative predictive value was 70 percent [93]. However, this model outperformed the three commercially available test kits for predicting fluoropyrimidine toxicity. Although two of the kits had 100 percent sensitivity (ie, identified all patients with severe toxicity), the positive predictive value was only 33 percent.

As with FU, routine testing for high-risk DPYD or TYMS alleles is not widely practiced prior to initiation of capecitabine because of the low frequency of finding a high-risk allele and the fact that patients who lack a high-risk variant may still suffer grade 3 or 4 FU-related toxicity. Nevertheless, testing is appropriate for patients who develop early severe toxicity (neutropenia, mucositis, diarrhea, neurotoxicity, and/or cardiotoxicity). (See 'DPD deficiency' above.)

Uridine triacetate is approved for emergency use in cases of capecitabine overdose or early-onset, severe, or life-threatening toxicity, such as might occur in a DPD-deficient patient. (See 'Uridine triacetate' above.)

Dosing — There appear to be large regional differences in the tolerance to capecitabine and other fluoropyrimidines [101]. These differences might, in part, be based on population-specific pharmacogenomic variability (eg, Asian patients seem to tolerate fluoropyrimidines better than non-Asians, and although not studied according to ethnicity, genetic factors that are associated with capecitabine sensitivity, such as SNPs, have been identified [122]). However, differences in lifestyle and diet (eg, dietary folate intake) could also contribute.

Because of these issues, the optimal dose of capecitabine, particularly for American patients, remains undefined. The initially approved dose for treatment of metastatic breast and colorectal cancer was 2500 mg/m² per day for 14 of every 21 days, but later studies suggest that this dose is too high, particularly in American patients. Lower doses (beginning at 2000 mg/m² per day for 14 of every 21 days) may improve the therapeutic index without compromising efficacy. (See "Systemic treatment of metastatic breast cancer in women: Chemotherapy", section on 'Capecitabine' and "Systemic chemotherapy for nonoperable metastatic colorectal cancer: Treatment recommendations", section on 'Oral fluoropyrimidines as substitutes for infusional FU'.)

Ftorafur — Two oral formulations of ftorafur (tegafur), an FU prodrug, have been developed and are in use in Japan; neither drug is available in the United States.

●UFT is a combination of ftorafur with uracil, which has been in widespread use in Japan for over 20 years. UFT is currently not available in the United States. Uracil competitively inhibits the enzyme DPD, leading to higher intratumoral concentrations of FU.

Grade 3 or 4 diarrhea is seen in up to 12 percent of patients treated with single-agent UFT and in 8 to 20 percent of those in whom UFT was given with LV [123,124]. Prompt discontinuation of UFT at the onset of diarrhea usually prevents severe GI toxicity.

●S-1 is an oral fluoropyrimidine that includes three different agents: ftorafur, gimeracil (5-chloro-2,4 dihydropyridine, a potent inhibitor of DPD), and oteracil (potassium oxonate, which inhibits phosphorylation of intestinal FU, thought responsible for treatment-related diarrhea).

In animal models, potassium oxonate is protective against FU-induced diarrhea [125,126]; in phase I and II clinical trials in Japan, the incidence of severe (grade 3 or 4) diarrhea with the S-1 formulation has been less than 10 percent [127-129]. In a randomized phase III Japanese trial of surgery followed by one year of S-1 chemotherapy versus surgery alone in gastric cancer, the incidence of grade 3 or 4 diarrhea was only 3.1 percent [130]. S-1 is also being evaluated in patients with advanced gastric cancer. (See "Systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer", section on 'Cisplatin plus a fluoropyrimidine'.)

Irinotecan — There are two types of diarrhea associated with irinotecan:

●Early-onset diarrhea with irinotecan occurs during or within several hours of drug infusion in 45 to 50 percent of patients and is cholinergically mediated (ie, related to increased motility) [14]. It is often accompanied by other symptoms of cholinergic excess, including abdominal cramping, rhinitis, lacrimation, and salivation. The mean duration of symptoms is 30 minutes; it is usually well controlled by subcutaneous or IV atropine. (See 'Altered intestinal motility' above.)

●In contrast, late irinotecan-associated diarrhea is not cholinergically mediated. The pathophysiology of late diarrhea appears to be multifactorial, with contributions from dysmotility and secretory factors, as well as a direct toxic effect on the intestinal mucosa [131,132].

Late diarrhea from irinotecan is unpredictable, noncumulative, and occurs at all dose levels. In early clinical trials of irinotecan, late diarrhea and neutropenia were the main dose-limiting toxicities [133,134]. Diarrhea of any grade was seen in 50 to 88 percent of patients, and it was severe in 9 to 31 percent. Diarrhea has been less common in later studies because of the stricter adherence to management guidelines (including routine early institution of high-dose loperamide) and the use of infusional rather than bolus FU in combination with irinotecan. (See "Management of acute chemotherapy-related diarrhea", section on 'Loperamide and diphenoxylate-atropine'.)

The median time to onset is approximately six days with the 350 mg/m² every three weeks schedule and 11 days with the weekly schedule (125 mg/m²) [132,135]. Late diarrhea is less common with the every three week schedule. In a randomized trial comparing the two administration schedules of single-agent irinotecan, the incidence of severe diarrhea was significantly less with the every three week schedule (36 versus 19 percent for weekly therapy) [136]. However, the incidence of cholinergic symptoms was significantly lower with weekly therapy (31 versus 61 percent). In some studies, older age, low performance status, and prior pelvic radiation were found to be predisposing factors [132]. For unclear reasons, diarrhea is more common in whites than in blacks receiving irinotecan-based therapy [137].

Irinotecan produces mucosal changes associated with apoptosis, such as epithelial vacuolization, and goblet cell hyperplasia, suggestive of mucin hypersecretion [2]. These changes appear related to the accumulation of the active metabolite of irinotecan, SN-38, in the intestinal mucosa [4].

SN-38 is glucuronidated in the liver and is then excreted in the bile. The conjugated metabolite SN-38G does not appear to cause diarrhea. However, SN-38G can be deconjugated in the intestines by beta-glucuronidase present in intestinal bacteria. A direct correlation has been noted between mucosal damage and either low glucuronidation rates or increased intestinal beta-glucuronidase activity [138-140]. Severe toxicity has been described with irinotecan in patients with Gilbert syndrome who have defective hepatic glucuronidation [139]. On the other hand, experimental studies have shown that inhibition of intestinal beta-glucuronidase activity with antibiotics protects against mucosal injury and ameliorates the diarrhea [140]. (See "Gilbert syndrome and unconjugated hyperbilirubinemia due to bilirubin overproduction" and "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'Irinotecan'.)

Common genetic polymorphisms of the UDP-glucuronyltransferase (UGT) gene can affect the metabolism of irinotecan. The possible impact of genetic variability on the toxicity of irinotecan is discussed below. (See 'Irinotecan plus FU' below.)

UGT1A1 polymorphisms — SN-38 is further metabolized by the polymorphic enzyme uridine diphospho-glucuronosyltransferase 1A1 (UGT1A1). In approximately 50 percent of the North American population, intratumoral enzymatic activity is reduced among those who inherit genetic polymorphisms such as the UGT1A1*28 (7/7 variant) allele. Approximately 10 percent of North American patients are homozygous for the UGT1A1*28 allele (which is the cause of Gilbert syndrome), and approximately 40 percent are heterozygous. In some studies, homozygotes (and to a lesser degree, heterozygotes) have had significantly higher rates of both diarrhea and myelosuppression with irinotecan. However, later analyses suggested that patients who inherit an UGT1A1*28 allele are mainly at risk for excess neutropenia and not diarrhea.

The manufacturer of irinotecan recommends that the starting dose be lowered in patients known to be homozygous for UGT1A1*28. However, they do not specifically recommend testing for the UGT1A1*28 variant prior to starting therapy, and testing prior to initiation of therapy is not widely practiced. This subject, as well as issues surrounding testing for the UGT1A1*28 allele, is discussed elsewhere. (See "Dosing of anticancer agents in adults", section on 'UGT1A1 polymorphisms and irinotecan'.)

Irinotecan plus FU — A standard regimen for treatment of metastatic colorectal cancer is the combination of irinotecan, FU, and LV [141,142]. Both irinotecan and FU have overlapping toxicity profiles; a major concern with early studies of this triplet regimen was the potential for enhanced GI toxicity. The spectrum of GI toxicity with combined irinotecan plus FU and LV is schedule-dependent:

IFL — In two United States Cooperative Group trials of bolus irinotecan plus weekly FU and LV (the IFL regimen), unacceptably high rates of early treatment-related mortality were noted [143-145]. In both trials, patients receiving irinotecan plus bolus FU and LV (either daily or weekly) had a threefold higher rate of treatment-related death than those enrolled on other arms [143]. Most of the early deaths appeared to be due to a cluster of mainly GI symptoms that included diarrhea, nausea, vomiting, and abdominal cramping, which was typically accompanied by dehydration, neutropenia, fever, and electrolyte abnormalities [144].

Later reports have suggested that 60-day mortality rates among patients initiating chemotherapy for metastatic colorectal cancer with the IFL regimen (1.3 to 4.4 percent) are no higher than those reported for bolus FU and LV (both by the Mayo regimen [daily administration of bolus FU and LV for five days each month; 6.7 percent] and the Roswell Park regimen [weekly bolus FU and high-dose LV; 7.6 percent]), and capecitabine monotherapy (5.7 percent) [146,147]. These later analyses likely reflect the increased understanding among oncologists of the risks associated with weekly bolus schedules of irinotecan plus FU and LV, and the need for more aggressive early treatment of GI toxicity when it occurs [144].

Specific therapeutic recommendations for the management of diarrhea (or abdominal cramping) in patients receiving IFL have been developed (table 4). However, use of this regimen has fallen out of favor, largely because of the risk of diarrhea and neutropenia.

FOLFIRI — GI toxicity is less severe with other irinotecan-containing regimens that utilize irinotecan plus LV and short-term infusional FU (eg, FOLFIRI). In at least four trials, rates of grade 3 or 4 diarrhea with FOLFIRI (every other week irinotecan plus short-term infusional FU and LV) were between 10 and 14 percent [142,148-150].

The better tolerability of regimens in which irinotecan is combined with infusional rather than bolus FU [148] has led to the widespread use of FOLFIRI rather than IFL. (See "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'Irinotecan'.)

Liposomal irinotecan plus fluorouracil and leucovorin — Liposomal irinotecan is a nanoliposomal encapsulated preparation that allows irinotecan to remain in circulation for a longer duration compared with standard irinotecan; this increases drug uptake within tumor cells and conversion of irinotecan to its active form, SN-38 [151]. Liposomal irinotecan is approved in combination with FU/LV for second-line treatment of gemcitabine-refractory metastatic pancreatic cancer based on results from the international phase III NAPOLI-1 trial [152]. Severe diarrhea, which can be of early-onset or late-onset type, occurred in 13 percent of those receiving combination therapy in this trial.

As with nonencapsulated irinotecan, the manufacturer of liposomal irinotecan recommends that the starting dose be lowered (from 70 to 50 mg/m² every two weeks) in patients homozygous for UGT1A1*28. However, they do not specifically recommend testing for the UGT1A1*28 variant prior to starting therapy. (See "Chemotherapy for advanced exocrine pancreatic cancer", section on 'Liposomal irinotecan'.)

Oxaliplatin combinations — Combinations of oxaliplatin plus FU and LV have become the most widely chosen first-line chemotherapy regimens for metastatic colorectal cancer, at least in North America. In addition, oxaliplatin-containing regimens have also been shown to provide a survival benefit over non-oxaliplatin-containing regimens for adjuvant therapy of stage III colon cancer. (See "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'First-line oxaliplatin plus FU/LV' and "Adjuvant therapy for resected stage III (node-positive) colon cancer", section on 'Oxaliplatin-based therapy'.)

Enterotoxicity with combined oxaliplatin and FU-containing chemotherapy is dependent on the schedule of FU administration. With oxaliplatin combined with short-term infusional FU (a regimen referred to as FOLFOX), rates of grade 3 or 4 diarrhea are less than 20 percent [153-156]. On the other hand, enterotoxicity is much more frequent and severe with regimens that combine oxaliplatin with weekly bolus FU and LV (eg, FLOX) [39], especially those that include daily bolus FU and LV [145]. (See "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'First-line oxaliplatin plus FU/LV'.)

This was shown in the NSABP C-07 trial, a comparison of the Roswell park regimen without or with (ie, FLOX) oxaliplatin as adjuvant therapy for stage II or III colon cancer [39]. During therapy, 79 of 1857 patients (4.3 percent) developed a syndrome of bowel wall injury characterized by hospitalization for management of severe diarrhea or dehydration and radiographic or endoscopic evidence of bowel wall thickening or ulceration, and the incidence was significantly higher in patients assigned to FLOX as compared with FU/LV (5.5 versus 3 percent). The incidence of bowel wall injury during chemotherapy was particularly high with FLOX as compared with FU and LV in patients aged 60 or older (6.7 versus 2.9 percent) and in females (9.1 versus 3.9 percent).

Enteric sepsis, characterized by grade 3 or worse diarrhea and grade 4 neutropenia (with or without bacteremia), occurred in 22 patients on FLOX and 8 patients on FU/LV. There were five deaths due to enteropathy, all in patients with enteric sepsis, with or without bowel wall injury. These results underscore the need to closely monitor patients treated with adjuvant FU/LV chemotherapy for diarrhea and provide aggressive management of this symptom complex, particularly if oxaliplatin has been added.

Capecitabine plus oxaliplatin — The combination of oxaliplatin and capecitabine (XELOX) has also been intensely investigated given its convenience. A phase III comparison of XELOX (capecitabine at 1000 mg/m² twice a day for 14 days plus oxaliplatin 130 mg/m² on day 1 every three weeks) with FOLFOX (continuous infusion of FU at 2250 mg/m² over 48 hours on days 1, 8, 15, 22, 29, and 36 plus oxaliplatin 85 mg/m² on days 1, 15, and 29 every six weeks) in patients with metastatic colorectal cancer reported a significantly lower rate of grade 3 or 4 diarrhea with XELOX (14 versus 24 percent) but a significantly higher rate of grade 1 or 2 hyperbilirubinemia (37 versus 21 percent) [157]. Similar rates of grade 3 or 4 diarrhea are reported in other trials as well. (See "Systemic chemotherapy for metastatic colorectal cancer: Completed clinical trials", section on 'Capecitabine plus oxaliplatin'.)

In patients with inoperable or metastatic cancers of the esophagus, gastroesophageal junction, or stomach, a phase III comparison of regimens containing epirubicin with either cisplatin or oxaliplatin and either FU or capecitabine reported a rate of grade 3/4 diarrhea of 12 percent for the combination of day 1 epirubicin 50 mg/m², day 1 oxaliplatin 130 mg/m² and twice daily capecitabine 625 mg/m² (EOX) [158]. (See "Systemic therapy for locally advanced unresectable and metastatic esophageal and gastric cancer".)

Pemetrexed — Pemetrexed is an antifolate with activity in non-small cell lung cancer and mesothelioma. In phase II and III trials of pemetrexed monotherapy, diarrhea (typically grade 1 or 2) has been reported in approximately 10 to 15 percent of patients [159-161]. (See "Systemic therapy for the initial management of advanced non-small cell lung cancer without a driver mutation" and "Systemic treatment for unresectable malignant pleural mesothelioma".)

Cabazitaxel — Cabazitaxel is a semisynthetic taxane that is approved for the treatment of advanced prostate cancer. Diarrhea is a frequent problem, developing in 15 to 50 percent of treated patients, but it is severe (grade 3 or 4) in only 1 to 6 percent [162-165]. Nevertheless, in a pivotal phase III trial conducted in men with advanced prostate cancer, some deaths occurred in men treated with cabazitaxel that were attributed to diarrhea and electrolyte imbalance [164]. (See "Chemotherapy in castration-resistant prostate cancer", section on 'Prior docetaxel'.)

Bortezomib and other proteasome inhibitors — Diarrhea is commonly seen with bortezomib, a proteasome inhibitor used in the treatment of multiple myeloma [166]. In the pivotal studies with this agent, diarrhea occurred in 51 percent of patients, with 8 percent of the events being grade 3 or 4. (See "Treatment of relapsed or refractory multiple myeloma", section on 'Proteasome inhibitors'.)

Diarrhea is less common and less severe with carfilzomib [167] and the orally active agent ixazomib [168].

Vorinostat, belinostat, and panobinostat — Vorinostat, a novel histone deacetylase inhibitor, was approved by the FDA in 2006 for the management of cutaneous T-cell lymphoma. In the pivotal studies of this agent, diarrhea was observed in 52 percent of the patients, with the great majority of these episodes being grade 1 or 2 events controllable by oral agents [169].

Belinostat is another histone deacetylase inhibitor that is approved for treatment of peripheral T-cell lymphoma. In an initial study, diarrhea was reported in 23 percent of treated patients but was severe in only 2 percent [170].

Panobinostat is another histone deacetylase inhibitor that is approved for refractory multiple myeloma. Diarrhea is frequent and is severe in approximately 25 percent of patients [171]. Dose modification guidelines are available in the United States prescribing information.

Lenalidomide — Lenalidomide can cause constipation or diarrhea. Among patients with diarrhea, bile salt malabsorption may be one potential etiology that responds to treatment including reduced fat intake and bile acid sequestrants [172].

Molecularly targeted agents — Diarrhea is a common side effect of several molecularly targeted agents.

Small molecule EGFR inhibitors — Diarrhea is common in patients receiving small molecule epidermal growth factor receptor (EGFR) TKIs, such as erlotinib, gefitinib, and afatinib [173]. For patients treated with gefitinib and erlotinib, diarrhea is most likely to occur within the first four weeks of treatment initiation; with afatinib, diarrhea is most likely to occur within the first seven days. Although diarrhea is reported in up to 90 percent of patients (especially those treated with afatinib [174]), it is severe in fewer than 15 percent and typically can be easily managed by the use of loperamide. Uncommonly, diarrhea necessitates dose reduction or treatment interruptions. Although diarrhea can have a profound effect on patients, the available evidence suggests that diarrhea may be a surrogate indicator of antitumor efficacy [175,176].

Synergistic toxicity may be a problem when these agents are combined with chemotherapy. Diarrhea has been a significant dose-limiting toxicity in a number of studies combining EGFR inhibitors with concurrent radiation and chemotherapy [177,178].

Small molecule inhibitors of VEGFR — Sorafenib, sunitinib, axitinib, regorafenib, ponatinib, pazopanib, cabozantinib, lenvatinib, and vandetanib are orally active inhibitors of multiple tyrosine kinases including the vascular endothelial growth factor receptor (VEGFR). Diarrhea is a prominent side effect of all VEGFR inhibitors. In clinical trials, diarrhea of any grade has been reported in 30 to 79 percent of patients (highest rates with vandetanib and lenvatinib), with severe diarrhea (grade 3 or 4) in 3 to 17 percent. (See "Toxicity of molecularly targeted antiangiogenic agents: Non-cardiovascular effects", section on 'Gastrointestinal toxicities'.)

BCR-ABL1 tyrosine kinase inhibitors — Imatinib, an inhibitor of BCR-Abl1 and other tyrosine kinases, such as KIT, causes diarrhea in approximately 30 percent of the patients, but severe diarrhea is rare [179]. Similarly, dasatinib, which targets BCR-ABL1, KIT, and the Src family of tyrosine kinase (among others), causes diarrhea in approximately 30 percent, which is severe in <5 percent. Hemorrhagic colitis is reported in patients treated with dasatinib, although the frequency with which this occurs is not established [180]. (See "Clinical use of tyrosine kinase inhibitors for chronic myeloid leukemia".)  

On the other hand, bosutinib, an inhibitor of BCR-ABL1 and the Src family of kinases, causes diarrhea in a higher proportion of patients (76 to 84 percent) and is severe in approximately 9 percent [181]. (See "Treatment of chronic myeloid leukemia in chronic phase after failure of initial therapy", section on 'Bosutinib'.)

Lapatinib and pertuzumab — Lapatinib, an orally active TKI that affects both the human epidermal growth factor receptor 2 (HER2; also called erbB-2) and EGFR (also called erbB-1), causes diarrhea in approximately 80 percent of patients; it is severe (grade 3 or 4) in 20 to 30 percent of patients [182]. Patients are typically managed with antidiarrheal agents; severe diarrhea may require hydration, electrolyte repletion, and/or interruption of therapy (recommended for grade 3 toxicity (table 1), and grade 1 or 2 toxicity that is complicated by moderate to severe abdominal cramping, grade 2 or worse nausea or vomiting (table 5), decreased performance status, sepsis, fever, neutropenia, bleeding, or dehydration). Treatment can be reintroduced at a lower dose when diarrhea resolves to grade 1 or less.

Pertuzumab is a recombinant monoclonal antibody directed against HER2, but it does not require HER2 overexpression for activity. Treatment-related diarrhea is frequent, but not commonly severe [183,184]. In a study of pertuzumab monotherapy in patients with metastatic breast cancer, diarrhea of any grade developed in 48 percent, but it was severe (grade 3 or 4) in only 3 percent [183].

Temsirolimus and everolimus — Temsirolimus and everolimus, which are inhibitors of the mammalian target of rapamycin (mTOR), can cause diarrhea, but severe GI toxicity is rare [185,186].

Anti-EGFR monoclonal antibodies

Cetuximab is a chimeric immunoglobulin G subclass 1 (IgG1) monoclonal antibody that binds to the extracellular domain of the EGFR, competitively inhibiting ligand binding. In contrast to small-molecule EGFR inhibitors, cetuximab-related diarrhea is generally not severe:

●A phase II study of cetuximab as monotherapy for 346 patients with metastatic colorectal cancer reported diarrhea of any grade in 12.7 percent [187].

●When toxicities were reported, regardless of attribution, to treatment in a phase II study of patients with lung cancer treated with cetuximab alone, 22.7 percent had diarrhea of any grade [188].

●Rates of grade 3 or 4 diarrhea in studies of single-agent cetuximab are only 1.5 to 2 percent [187-190].

Panitumumab is a fully human IgG2 monoclonal antibody directed against the EGFR. A phase III comparison of best supportive care (BSC) with or without panitumumab reported diarrhea of any grade in 21 percent of patients receiving panitumumab (grade 3, 1 percent), compared with 11 percent with BSC alone (none grade 3) [191]. Similar results are reported by others with panitumumab monotherapy [192].

ALK inhibitors — Crizotinib and ceritinib are orally active inhibitors of the anaplastic lymphoma kinase (ALK); both are approved for treatment of advanced or metastatic non-small cell lung cancer if the tumor contains a characteristic EML4-ALK fusion oncogene. (See "Anaplastic lymphoma kinase (ALK) fusion oncogene positive non-small cell lung cancer", section on 'Crizotinib' and "Anaplastic lymphoma kinase (ALK) fusion oncogene positive non-small cell lung cancer", section on 'Ceritinib'.)

Diarrhea is a common side effect, particularly with ceritinib, but rarely severe:

●In a review of over 250 treated patients, the incidence of all-grade diarrhea with ceritinib was 86 percent; it was severe (≥grade 3) in 6 percent.

●In a review of 397 patients treated with crizotinib, 49 percent developed diarrhea of any grade, and it was severe in <1 percent.

MEK inhibitors — Trametinib and cobimetinib are orally active inhibitors of the mitogen-activated protein kinase enzymes MEK1/MEK2; they are both approved for treatment of metastatic melanoma with a specific BRAF^V600 mutation. (See "Molecularly targeted therapy for metastatic melanoma", section on 'MEK inhibition'.)

Diarrhea is a frequent complication of these drugs. In clinical trials, approximately one-half of patients develop diarrhea, but it is severe (grade 3 or worse) in fewer than 5 percent of cases [193-195]. The United States prescribing information for both drugs recommends withholding the drug for grade 2 or worse adverse reactions, including diarrhea (table 1), and permanent discontinuation if toxicity does not improve to grade 0-1 within three or four weeks or for any recurrent grade 4 toxicity.

BTK inhibitors — Ibrutinib and acalabrutinib are orally active inhibitors of Bruton tyrosine kinase (BTK), a mediator of the B-cell receptor signaling pathway that inhibits malignant B-cell survival; both are approved as a single agent for treatment of mantle cell lymphoma, and ibrutinib is also used in chronic lymphocytic leukemia. (See "Treatment of relapsed or refractory mantle cell lymphoma", section on 'Ibrutinib' and "Treatment of relapsed or refractory mantle cell lymphoma", section on 'Acalabrutinib' and "Treatment of relapsed or refractory chronic lymphocytic leukemia", section on 'Ibrutinib'.)

In a phase III trial, the incidence of treatment-related diarrhea with ibrutinib was 42 percent; only 4 percent of patients had severe (grade 3 or 4) diarrhea (table 1) [196]. The FDA-approved manufacturer's labeling provides specific guidelines for dose reduction in response to adverse events, including diarrhea.

The risk of diarrhea was slightly less (all-grade 30 percent, 3 percent grade 3 or worse) in a phase II trial of acalabrutinib [197]. The United States prescribing information provides specific guidelines for dose reduction in the setting of grade 3 or higher nonhematologic toxicity.

Rituximab — Rituximab, an anti-CD20 monoclonal antibody used to treat B-cell lymphoma, can cause new-onset ulcerative colitis or exacerbation of preexisting colitis [9].

Idelalisib — Idelalisib is an oral inhibitor of phosphoinositide 3-kinase (PI3K) delta; it is approved for treatment of relapsed chronic lymphocytic leukemia, follicular lymphoma, and small lymphocytic lymphoma. (See "Treatment of relapsed or refractory chronic lymphocytic leukemia", section on 'Idelalisib'.)

Across clinical trials, approximately 14 percent of treated patients have developed severe diarrhea or colitis (grade 3 or worse) [198-200]. Diarrhea can occur at any time during treatment, and it responds poorly to antimotility agents. Following interruption of therapy and, in some cases, the use of glucocorticoids or budesonide [199], the median time to resolution is between one and four weeks. Some patients with moderate to severe diarrhea have developed serious and fatal intestinal perforation. (See 'Idelalisib' below.)

Abemaciclib — Abemaciclib is an inhibitor of cyclin-dependent kinases (CDK) 4 and 6; it is approved for treatment of hormone receptor-positive metastatic breast cancer. (See "Treatment approach to metastatic hormone receptor-positive, HER2-negative breast cancer: Endocrine therapy and targeted agents", section on 'Aromatase inhibitors plus CDK 4/6 inhibitors'.)

Diarrhea is a frequent adverse reaction, occurring in 81 to 86 percent of treated patients, but it is typically low grade (severe in 10 to 13 percent) [201,202]. Diarrhea typically responds to antidiarrhea therapy, and treatment discontinuation is rarely needed.

CONSTIPATION — Constipation can be defined as a decreased frequency of defecation (usually less than three bowel movements per week) accompanied by discomfort or difficulty. It is a common problem in patients with cancer, usually being due to a combination of poor oral intake and drugs, such as opioid analgesics or antiemetic agents, that slow intestinal transit time. As an example, 5HT3 receptors are present on enteric neurons, and ondansetron was shown to slow colonic transit time in healthy subjects [203].

Methylnaltrexone, a pure opiate antagonist that does not cross the blood brain barrier, may be useful to treat constipation when opioid analgesics are a significant causal factor for the constipation. Methylnaltrexone does not block the analgesic activity of opiates or precipitate opiate withdrawal. In a randomized study, subcutaneous methylnaltrexone administration showed excellent efficacy and did not affect central analgesia or precipitate opioid withdrawal [204]. As a result, methylnaltrexone received US Food and Drug Administration (FDA) approval in 2008 for the management of opioid-induced constipation in patients with advanced illness who are receiving palliative care. Its use in other settings, such as medication-induced constipation, has not been carefully investigated yet. (See "Prevention and management of side effects in patients receiving opioids for chronic pain", section on 'Opioid bowel dysfunction'.)

Specific drugs

Vinca alkaloids — Constipation is rarely a dose-limiting toxicity for chemotherapeutic agents, except for the vinca alkaloids (eg, vincristine, vinblastine, and vinorelbine), especially vincristine (see "Overview of neurologic complications of non-platinum cancer chemotherapy", section on 'Vincristine').

These drugs have pronounced neuropathic effects and increase gastrointestinal transit time [205]. The constipating effect of vinca alkaloid therapy is usually apparent after the first dose and is typically not cumulative. It is most prominent 3 to 10 days after chemotherapy and then resolves, in most cases, after a few days [206].

Constipation occurs in one-quarter to one-third of patients [207,208] and is severe (table 6) in 2 to 3 percent [207,209]. In one series of 392 patients, 2.8 percent required hospitalization for adynamic ileus [207].

Vincristine-induced constipation is more severe at higher doses (above 2 mg). This was illustrated in a report of 104 patients with Hodgkin or non-Hodgkin lymphoma [210]. Vincristine was given in a non-capped dose of 1.4 mg/m², and 90 percent of patients received more than 2 mg in the first dose. Severe constipation occurred in 10 percent. Rapid improvement usually occurred within a few weeks after the cessation of therapy.

Constipation is also more frequent and may be more severe (presenting as a bowel obstruction or paralytic ileus) in patients treated with the liposome-encapsulated version of vincristine (Marqibo), which is approved only for treatment of refractory adult acute lymphoblastic leukemia. (See "Treatment of relapsed or refractory acute lymphoblastic leukemia in adults", section on 'Liposomal vincristine'.)

Thalidomide and analogs — Thalidomide and its analogs lenalidomide and pomalidomide have shown promise for the treatment of refractory multiple myeloma and other disorders. (See "Treatment of relapsed or refractory multiple myeloma".)

The most common toxicity with thalidomide, beside sedation, is constipation. In a major myeloma trial, constipation developed in 35 percent of patients at 200 mg/day and in 59 percent at 800 mg/day [211]. These rates seem higher than those observed in a phase II trial of thalidomide (starting dose 800 mg daily) in patients with recurrent high-grade glioma, in which constipation was the most common toxicity but only occurred in 19 percent of patients; no severe episodes were noted [212].

Thalidomide-induced constipation is dose-dependent, appears early after the initiation of therapy (within two to four days) in most patients, and is more severe in older adults and in those receiving opioid analgesics [213]. The mechanism might be neuromuscular inertia with resultant hypotonia. Thalidomide-induced hypothyroidism should be considered in patients with persistent constipation or if it appears late in the disease course.

Lenalidomide, an immunomodulatory agent derived from thalidomide, is also associated with constipation, although it is rarely severe [214-216].

In clinical trials, pomalidomide has been associated with diarrhea or constipation in approximately one-third of treated patients, none of which were severe [217].

Vandetanib — Vandetanib is a multitargeted inhibitor of several tyrosine kinases. In clinical trials involving patients with medullary thyroid cancer and lung cancer, vandetanib was associated with constipation in 9 to 37 percent of patients, with 0 to 3 percent severe [218-220]. Vandetanib is more often associated with diarrhea. (See 'Small molecule inhibitors of VEGFR' above.)

Belinostat — Belinostat is a histone deacetylase inhibitor that is approved for treatment of peripheral T-cell lymphoma. In an initial clinical trial involving 129 patients, belinostat was associated with constipation in 23 percent of patients, with 2 percent severe [170]. Diarrhea was equally common. (See 'Vorinostat, belinostat, and panobinostat' above.)

Treatment — The treatment of chemotherapy-induced constipation begins with anticipation and prevention. Laxatives should be started at the first sign of constipation or should be given routinely to prevent constipation. The most frequently used laxatives are docusate, senna, or bisacodyl. If these agents are not effective, magnesium salts, polyethylene glycol, lactulose, or sorbitol are often effective [221]. (See "Management of chronic constipation in adults".)

INTESTINAL PERFORATION — Bowel perforation is an uncommonly encountered complication that seems to be associated with antiangiogenic agents, particularly bevacizumab, a monoclonal antibody targeting the vascular endothelial growth factor (VEGF).

This complication is not unique to agents that target angiogenesis. Bowel perforation has also been seen (albeit rarely) with other tumors involving the gastrointestinal (GI) tract that respond rapidly to conventional cytotoxic chemotherapy (eg, GI tract lymphomas). (See "Treatment of extranodal marginal zone lymphoma of mucosa associated lymphoid tissue (MALT lymphoma)".)

Angiogenesis inhibitors — All VEGF-targeted therapies, including bevacizumab, aflibercept, and the orally active antiangiogenic tyrosine kinase inhibitors, can cause gastrointestinal perforation (GIP), although this complication is best described in patients receiving bevacizumab. GIP has been reported in patients treated with bevacizumab for a variety of malignancies but is most often described in the setting of metastatic colorectal cancer and epithelial ovarian cancer. GIP can occur anywhere along the GI tract, often away from tumor sites or prior surgical anastamoses. (See "Toxicity of molecularly targeted antiangiogenic agents: Non-cardiovascular effects", section on 'Intestinal perforation/fistula formation'.)

Although several risk factors have been described for GIP during bevacizumab treatment, bowel perforation may occur even in the absence of predisposing risk factors, and it remains difficult to predict which patients will develop this complication. Many cases involve perforation of an in situ bowel primary. However, GIP can also occur at previously resected primary sites, often in the setting of previous irradiation or a prior anastomotic leak. GIP can also occur during bevacizumab treatment of malignancies that lack disease within the peritoneal cavity (eg, primary malignant brain tumors).

In order to minimize the risk of GIP and fistula formation, at least 28 days (preferably six to eight weeks) should elapse between surgery and the last dose of bevacizumab, except in emergency situations. Clinicians should maintain a high index of suspicion for GIP in patients who develop acute abdominal pain while receiving bevacizumab, even if they have no apparent risk factors.

Perforation may be asymptomatic, or it can present with abdominal pain from peritoneal contamination, free air, hemoperitoneum, or an intraabdominal abscess. Patients with confirmed or highly suspected GIP whose overall condition is unstable secondary to the GIP should be considered for immediate surgical repair or diversion. Those who are more stable can be considered for less invasive management strategies such as bowel rest and broad-spectrum antibiotics with or without percutaneous drainage of concurrent abscesses. The timing of the presentation, the patient's overall condition, their goals and wishes, and overall prognosis are important factors in the decision to explore these patients surgically. This subject is addressed in detail elsewhere. (See "Toxicity of molecularly targeted antiangiogenic agents: Non-cardiovascular effects", section on 'Intestinal perforation/fistula formation'.)

Thalidomide in multiple myeloma — At least one case report describes four cases of bowel perforation in patients receiving thalidomide for multiple myeloma [222]. The true incidence is unknown.

Erlotinib — Cases of GIP (some fatal) have also been reported in patients receiving the small-molecule epidermal growth factor receptor (EGFR) inhibitor erlotinib [223]. (See "Systemic therapy for advanced non-small cell lung cancer with an activating mutation in the epidermal growth factor receptor", section on 'EGFR TKI toxicity'.)

Idelalisib — Fatal and serious intestinal perforation has occurred in patients treated with idelalisib, an oral inhibitor of phosphoinositide 3-kinase (PI3K) delta that is approved for treatment of relapsed chronic lymphocytic leukemia, follicular lymphoma, and small lymphocytic lymphoma. At the time of perforation, some patients have had moderate to severe diarrhea. Idelalisib should be discontinued permanently in patients who experience intestinal perforation. (See 'Idelalisib' above.)

Trametinib — The MEK inhibitor trametinib has been associated with GIP. Across clinical trials of trametinib administered as a single agent or in combination with dabrafenib, the risk of GIP was 0.3 percent [224].

SUMMARY AND RECOMMENDATIONS — Gastrointestinal (GI) toxicity due to chemotherapeutic drugs is a common problem in cancer patients.

●Acute chemotherapy-related diarrhea (CRD) is most commonly described with fluoropyrimidines (particularly fluorouracil [FU] and capecitabine), irinotecan, pemetrexed, cabazitaxel, bortezomib, vorinostat, and several molecularly targeted agents (including sorafenib, sunitinib, afatinib, ceritinib, and agents targeting the epidermal growth factor receptor [EGFR], including lapatinib and erlotinib). (See 'Diarrhea' above.)

CRD occurs through three major different mechanisms: increased secretion of electrolytes caused by luminal secretagogues or reduced absorptive capacity (due to surgery or epithelial damage), called secretory diarrhea; increased intraluminal osmotic substances leading to osmotic diarrhea; or altered GI motility. In addition, patients receiving agents targeting the vascular endothelial growth factor receptor (VEGFR) may develop direct ischemic mucosal damage, and patients receiving immune checkpoint inhibitors may develop an immune-mediated colitis. (See 'Pathogenesis/mechanisms' above.)

In addition, specific colitis syndromes that may arise in patients receiving chemotherapy include neutropenic enterocolitis, ischemic colitis, and clostridium difficile-associated colitis. (See 'Colitis syndrome that may arise in patients treated with chemotherapy' above.)

Finally, patients who develop diarrhea during chemotherapy may also suffer from organic causes of diarrhea, such as bacterial overgrowth, fat or bile acid malabsorption, intake of excess quantities of sorbitol or lactose intolerance, and inflammatory and infectious causes, which should not be overlooked. (See 'Differential diagnosis' above.)

●CRD typically begins with an increasing frequency of bowel movements and/or a loosening of the stool consistency. Excessive gas and/or intestinal cramping commonly accompanies CRD. As CRD progresses, it can become severe, with frequent watery stools. The severity of CRD is often described using the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) grades; the latest version is outlined in the table (table 1). (See 'Clinical manifestations' above.)

CRD can be debilitating and, in some cases, life-threatening. Given the risk for dehydration and infection, severe CRD frequently requires hospital admission for adequate supportive care. Management of CRD according to the severity of the diarrhea and the presence or absence of other risk factors is outlined in the algorithm (algorithm 1) and presented in detail elsewhere. (See "Management of acute chemotherapy-related diarrhea".)

●Diarrhea is particularly frequent in patients treated with fluoropyrimidines, such as FU and its oral derivative capecitabine. Both the therapeutic efficacy and the frequency of diarrhea associated with FU are increased when given concurrently with leucovorin (LV). The highest frequency of diarrhea occurs with bolus rather than continuous 24-hour infusion of FU/LV for up to five days, particularly with weekly administration, but it occurs with all schedules. (See 'Fluorouracil' above.)

Dihydropyrimidine dehydrogenase (DPD) is the rate-limiting enzyme in fluoropyrimidine catabolism. Enzyme activity varies widely, with most of the variability arising from genetic polymorphisms in the DPD gene (DPYD). Patients who are partially or totally deficient in DPD activity cannot adequately degrade fluoropyrimidines, leading to an increased risk of severe, sometimes fatal, toxicity, including diarrhea, myelosuppression, and mucositis. In addition to DPYD, high-risk polymorphisms in the thymidylate synthetase gene (TYMS) may be associated with an increased risk of severe toxicity from fluoropyrimidine-based chemotherapy (table 2), although the data are less certain than with DPYD. (See 'Predictive markers' above.)

Testing is commercially available that might identify patients who are at risk for severe toxicity from fluoropyrimidines based upon DPYD and TYMS genotype. Some authors suggest that testing should ideally be carried out before initiation of treatment to identify high-risk patients; however, this is a controversial area, and this approach has not been widely adopted. Although genotyping may identify a small fraction of patients for whom serious toxicity is a concern, most patients who develop serious toxicity to fluoropyrimidines do not have mutations in either DPYD or TYMS (ie, sensitivity is limited). Furthermore, the frequency of high-risk alleles in the general population is quite small (mostly <1 percent), and many patients who inherit these polymorphisms will not suffer from undue toxicity (ie, the positive predictive value has been variable). For these reasons, we generally reserve genotyping for those patients who have unexpected toxicity (myelosuppression, mucositis, diarrhea, neurotoxicity, cardiotoxicity) during the first few cycles of fluoropyrimidine therapy. (See 'Pharmacogenetic testing for DPYD and TYMS variants' above.)

●Late diarrhea from irinotecan is unpredictable, noncumulative, and occurs at all dose levels. In early clinical trials of irinotecan, late diarrhea and neutropenia were the main dose-limiting toxicities. Diarrhea has been less common in later studies because of the stricter adherence to management guidelines (including routine early institution of high-dose loperamide) and the use of infusional rather than bolus FU in combination with irinotecan.

The active form of irinotecan, SN-38, is further metabolized by the polymorphic enzyme uridine diphospho-glucuronosyltransferase 1A1 (UGT1A1). Approximately 10 percent of North American patients are homozygous for the UGT1A1*28 allele (which is the cause of Gilbert syndrome). In some studies, homozygotes (and to a lesser degree, heterozygotes) have had significantly higher rates of both diarrhea and myelosuppression with irinotecan. However, later analyses suggested that patients who inherit an UGT1A1*28 allele are mainly at risk for excess neutropenia and not diarrhea. Nevertheless, the manufacturer of irinotecan recommends that the starting dose be lowered in patients known to be homozygous for UGT1A1*28. (See 'UGT1A1 polymorphisms' above.)

●Several molecularly targeted agents have been associated with CRD; the most common are small-molecule tyrosine kinase inhibitors targeting the EGFR, especially afatinib and lapatinib, anaplastic lymphoma kinase (ALK) inhibitors such as ceritinib, and monoclonal antibodies directed against the EGFR. (See 'Molecularly targeted agents' above.)

●Constipation is rarely a dose-limiting toxicity for chemotherapeutic agents, except for the vinca alkaloids, especially vincristine. (See 'Vinca alkaloids' above.)

Treatment should focus on anticipation and prevention. Laxatives should be started at the first sign of constipation or should be given routinely to prevent constipation. (See 'Constipation' above.)

●Bowel perforation is an uncommonly encountered complication that seems to be associated with antiangiogenic agents, particularly bevacizumab, a monoclonal antibody targeting the VEGF. (See 'Intestinal perforation' above.)

This complication is not unique to agents that target angiogenesis. Bowel perforation has also been seen (albeit rarely) with other tumors involving the GI tract that respond rapidly to conventional cytotoxic chemotherapy (eg, GI tract lymphomas). (See "Treatment of extranodal marginal zone lymphoma of mucosa associated lymphoid tissue (MALT lymphoma)".)