Chemotherapy

Cytotoxic chemotherapy agents have a direct effect on the GI mucosa causing inflammation,11–13 oedema, ulceration and atrophy. Increased bowel permeability combined with the secondary effects of immunosuppression predispose to increased susceptibility to GI transmural infection potentially leading to septicaemia, shock, associated hypotension and secondary mucosal ischaemia.

New-onset lactose intolerance is the cause of diarrhoea and bloating during 5-fluorouracil chemotherapy in 10% of patients.14 15 Clinical experience suggests that small bowel bacterial overgrowth, bile acid malabsorption and pancreatic insufficiency are three other important contributory factors to chemotherapy-induced GI symptoms.16

Hepatic side effects include reactivation of hepatitis B,17 severe sinusoidal obstruction syndrome (especially oxaliplatin-based regimens)18 and steatosis. Following treatment with irinotecan in particular, steatosis may progress to steatohepatitis.19 This may be exacerbated by pre-existing conditions (eg, alcohol excess, diabetes and obesity). Both steatosis and steatohepatitis are potentially reversible after cessation of chemotherapy but occasionally contribute to morbidity and mortality.18–21 The detailed management of these conditions is described elsewhere.22

Chronic GI side effects of chemotherapy have not been studied systematically or prospectively. Clinical experience suggests that a small proportion of patients do have ongoing GI problems with constipation, diarrhoea, flatulence, bloating and pain, and that small bowel bacterial overgrowth is a frequent cause of these symptoms. However, specific drugs, their cumulative dose, the degree of immunosuppression during treatment together with the degree of damage to the mucosa, submucosa and GI stem cells may also play a role in the development of chronic problems.16 23 The interaction with other cancer treatments complicates the clinical picture further.

Biological agents

Rapidly increasing numbers of biological agents are being introduced for cancer therapy. This includes both immunotherapy and inhibitors of specific molecular targets. The main categories of targeted therapies currently include tyrosine kinase inhibitors (eg, erlotinib, imatinib, gefitinib, sorafenib), proteasome inhibitors (eg, velcade) and anti-angiogenesis agents (eg, bevacizumab). The spectrum of GI toxicity with these agents and their causes are poorly defined. Biological agents have multiple interactions with conventional drugs and, when new drugs are prescribed, the potential for interactions must always be checked. It is particularly important for gastroenterologists to know that altering gastric pH in a patient taking a biological agent orally can markedly affect its bioavailability. In the emergency setting, if there is any possibility that the biological agent is the cause of severe symptoms, it is always acceptable that the agent should be stopped while waiting for urgent advice from the oncologist treating the patient. Acute severe GI symptoms should otherwise be managed normally.

Radiotherapy

Radiotherapy initially causes mucosal changes characterised by inflammation or cell death, but subsequently persistent cytokine activation in the submucosa leads to progressive ischaemia, fibrosis and loss of stem cells.24 These ischaemic and fibrotic changes potentially cause impairment of GI physiological function(s). Chemotherapy increases the sensitivity of non-cancerous tissues to damage from radiotherapy. Chronic GI dysfunction may follow without pause from acute symptoms induced by radiotherapy or may arise de novo months, years or even decades later. The time allowed for follow-up in most studies prevents the recognition of these late side effects, their frequency or severity.25

It is argued that introduction of new chemotherapy and biological agents and more targeted radiotherapy techniques over the last two decades will diminish toxicity rates,26 27 but the long-term effects of these new cancer treatments are unknown.28 For example, the technique of intensity-modulated radiotherapy (IMRT) has been introduced widely. One possibly important consequence of IMRT is that more organs are exposed to low-dose irradiation than was the case with conventional treatment. It will be some time before definitive evidence from clinical trials shows the true impact of this on acute and chronic toxicity. Clinical experience suggests that IMRT simply changes the timing and spectrum of toxicity. There are virtually no data on the long-term effects of even newer techniques such as proton beam therapy and cyber knife treatment.

Surgery

Radical resectional surgery may cause significant disruption of GI physiology. This includes disturbance in intestinal transit, altered gastric emptying, enzymatic digestion and malabsorption reflecting anatomical disruption and stasis, bacterial overgrowth, altered bile acid secretion and absorption and hepatic insufficiency. Hepatopancreatobiliary resections carry the inherent risk of subsequent biliary strictures (which may be due either to benign fibrosis at the anastomosis or disease recurrence) resulting in obstructive jaundice. In the past, many patients had limited survival after primary cancer surgery but symptom complexes which were common—for example, after upper GI surgery for peptic ulceration—are now being observed in long-term cancer survivors, although many clinicians will no longer be familiar with these.

Non-resectional ablation techniques

Radiofrequency or microwave ablation and tumour embolisation with a variety of agents including radioactive beads are being used increasingly with both palliative and curative intent to treat liver tumours. Complications include bleeding, ulceration, ischaemia or perforation of adjacent bowel, abscess formation, hepatic artery aneurysm and tumour track seeding.

Infection

Neutropenic sepsis is a common complication of cancer chemotherapy precipitating GI symptoms, which usually respond quickly to antibiotics. In patients with worsening or severe diarrhoea, one single stool specimen is sufficient for the detection of bacteria or toxins, however three separate specimens are required to exclude parasitological causes with sufficient diagnostic sensitivity. Early endoscopic assessment is also mandatory as stool culture may not detect viral infection, toxin-negative Clostridium difficile or drug-induced colitis. Endoscopy in a neutropenic patient predisposes to sepsis, although the degree of the increased risk is unclear. Febrile neutropenic patients should already be on antibiotic therapy. Recent British Society of Gastroenterology guidelines (grade C evidence: expert opinion) suggest that afebrile patients with a neutrophil count <0.5×10⁹/l need to be offered antibiotic prophylaxis only for GI endoscopic procedures associated with high risk of bacteraemia such as variceal sclerotherapy, oesophageal dilation, laser therapy and endoscopic retrograde cholangiopancreatography with biliary obstruction. Gram-negative aerobic (and, less frequently, anaerobic) bacteria including Escherichia coli are the most likely pathogens in these conditions, and the choice of prophylactic antibiotics should reflect the local sensitivities of organisms.30

The two most important treatable pathogens which regularly cause severe morbidity or death in patients receiving treatment for cancer are cytomegalovirus (CMV) and C difficile (10% are toxin-negative). The type of endoscopic assessment which best identifies stool culture-negative pathogens is not clearly defined. However, upper GI endoscopy with duodenal biopsies and aspirate and flexible sigmoidoscopy with left colonic biopsies seem to produce results equivalent to full colonoscopy and ileal biopsy16 31 32 while avoiding the need for full bowel preparation and reducing risk.

CMV infection may affect the whole GI tract but is most commonly found in the oesophagus and colon. Common symptoms include diarrhoea (up to 80%), bleeding (up to 64%), fever (up to 50%) and abdominal pain (19–50%). Endoscopy may demonstrate the presence of multiple ulcers. Serology, viral culture and PCR techniques are not reliably positive for 3 or 4 weeks after the onset of symptoms. Earlier diagnosis may be available using the newer shell vial assay and from examination of endoscopic biopsies. Biopsies should be taken both from the centre/base of ulcers (site of highest yield for CMV) and from the edge (which gives a higher yield in herpes simplex virus infection).33 34 In a sick immunosuppressed patient with relevant symptoms, early empirical treatment with ganciclovir should be considered.

The typical endoscopic appearance of C difficile at flexible sigmoidoscopy is often diagnostic in toxin-negative patients. However, pseudomembrane formation requires neutrophil involvement, and the typical macroscopic and microscopic appearance may be altered or be completely absent in neutropenic patients.35 Immunosuppressed patients with C difficile are at high risk of early progression to fulminant toxic megacolon, so delay in investigation and treatment is potentially dangerous.

Many other pathogens including amoebae, giardia, viruses such as herpes simplex virus, rotavirus or adenovirus, bacterial pathogens and fungi may be responsible for symptoms. Recurrent infections with different organisms in immunosuppressed patients may mandate repeated endoscopic reassessments at short intervals. More than one pathogen may be responsible.

Anorectal sepsis in neutropenic patients is a frequently forgotten cause of morbidity. Clinical assessment by an experienced colorectal surgeon supplemented by MRI scanning can often be helpful in detecting an occult site for recurrent infection.

Typhlitis and neutropenic enterocolitis

Typhlitis and neutropenic enterocolitis carry a high mortality rate because of the risk of rapid progression to ischaemia, necrosis, haemorrhage, perforation and multisystem organ failure. Typhlitis follows chemotherapy-induced neutropenia and is characterised by inflammation localised to the caecal wall, possibly caused by bacterial invasion. If other parts of the GI tract are involved (eg, the terminal ileal wall or elsewhere), the term ‘neutropenic enterocolitis’ is more appropriate. Bowel wall thickening with or without dilation is usually seen on cross-sectional imaging. Clinical features include fever, abdominal pain, nausea, vomiting and diarrhoea. Colonoscopy is contraindicated as it carries a very high risk of perforation.

Data on the optimal management are limited. Bowel rest, intravenous fluids, parenteral nutrition, broad spectrum antibiotics and normalisation of neutrophil counts are usually recommended.36 When inflammation is limited to the caecum and terminal ileum, clinical experience suggests that most patients can be managed conservatively. The presence of a local mass needs repeated imaging to exclude abscess formation or perforation. Perforation, persistent GI bleeding and clinical deterioration may mandate surgical intervention. As in the management of any acute severe colitis, frequent clinical reassessment and early surgical consultation is advised.

Haemorrhage

Causes of bleeding include chemotherapy-induced ischaemia (particularly induced by taxanes13 37 and bevacizumab38), infections (particularly CMV and Candida), graft versus host disease (which can occur after stem cell transplantation when the newly transplanted material attacks the transplant recipient's body), autoimmune colitis after treatment with ipilimumab,11 acute radiotherapy-induced ulceration, drug-or radiotherapy-induced inflammatory bowel disease, neutropenic enterocolitis and oxaliplatin-induced portal hypertension.39 Patients should be managed like any other high-risk GI bleed. Experienced endoscopists must be involved from the onset. Although endoscopy may be more hazardous, early endoscopic therapy may also be life-saving.

Thrombocytopenia is common in patients with cancer undergoing chemotherapy. Endoscopists should be aware that, even with apparently adequate platelet counts, chemotherapy-induced platelet dysfunction may affect normal homeostatic mechanisms. There are no robust data as to the minimal safe platelet count for safe therapeutic endoscopy. Endoscopists should consider ensuring in advance that platelet support is available should it be required when performing therapeutic procedures in patients with a platelet count below 50–80 000/μl.

The endoscopic appearance may not be diagnostic without biopsy. However, even flexible endoscopic biopsy under direct vision can be hazardous in the presence of thrombocytopenia, incipient ischaemic necrosis or previous radiotherapy, especially brachytherapy. Endoscopic intervention may be sufficient for bleeding from discrete sites, but interventional radiology with embolisation or surgery may be required in extensive mucosal change. Hyperbaric oxygen therapy is probably the treatment of choice for radiotherapy-induced bleeding occurring at multiple sites throughout the small and/or large bowel.40–42

Perforation

Perforation may result from spontaneous tumour necrosis, which may or may not be due to chemotherapy or progression of cytotoxic drug-induced ulceration, often on a background of adjunctive corticosteroid or non-steroidal anti-inflammatory drug treatment. The antivascular endothelial growth factor monoclonal antibody bevacizumab causes ulceration, fistulation or free perforation in 0.9% of patients within 1 year of treatment.43 This may be at the primary tumour site but also within colonic diverticula or otherwise normal areas of stomach and duodenum. Bevacizumab may increase the risk of stent-related44 and spontaneous perforation and is associated with bleeding, poor wound healing and thromboembolism. Two tyrosine kinase inhibitors, erlotinib and gefitinib, are also associated with bowel perforation. Surgical treatment is essential as long as the patient is fit enough, and therapeutic resection may be the best approach if the primary tumour has perforated. If circumstances permit, referral to a specialist surgeon is indicated.

Mesenteric ischaemia and infarction

Spontaneous mesenteric vascular insufficiency can be induced by the hypercoagulable state associated with some cytotoxic agents.45 This can affect both diseased and unaffected small bowel. The mortality rate is high and a high degree of suspicion is needed to diagnose ischaemia. The aetiology may be venous or arterial and expert radiology may help in assessment and management. Optimal management of acute intestinal ischaemia requires early assessment by an experienced surgical team. The options—depending on the general state of the patient—include full anticoagulation if the bowel is viable, through to staged resection, often requiring repeat laparotomies and open abdomen techniques.

Chemotherapy-associated mesenteric ischaemia can present with acute abdominal pain, but also can produce small bowel strictures causing small bowel obstruction. It must be treated by a combination of nutritional support, repeated clinical assessment by experienced surgeons and appropriate anticoagulation.

Hepatic veno-occlusive disease/portal vein thrombosis

This is a very frequent cause of early mortality among patients receiving high-dose chemoradiation or stem cell transplantation.46 Activation of the coagulation cascade and inflammatory processes following endothelial injury results in a hypercoagulable state. The possibility must always be considered in patients presenting with jaundice, pain or ascites. Many patients will, however, have rather non-specific symptoms or biochemical changes and early CT scanning with contrast may be diagnostic. Early anticoagulation may be life-saving.

Bowel obstruction

Obstruction usually affects the small bowel or, after pelvic radiotherapy, the sigmoid. Several factors may contribute in individual patients. It may develop as a result of benign causes such as changes in intestinal transit, medical causes (see below), adhesions or radiotherapy-induced fibrosis, or malignant causes such as recurrent cancer or peritoneal carcinomatosis.

Incidence and prevalence figures

GI side effects are underestimated in the literature and within clinical trials.56–58 Case notes frequently do not record side effects except when patients require surgery as part of their management.59 60 When prospective data are available,3 they are invariably based on symptom questionnaires rather than objective markers. Many current questionnaires are inadequately sensitive, do not use reproducible methodology and ignore issues important for patients—for example, severe flatulence or urgency of defaecation.49 61–65 Focusing on symptoms without confirmatory objective investigations is also potentially misleading. Common bowel disorders produce identical symptoms to those arising as side effects of cancer therapies.3 49 66 There is an urgent need for better tools which can be applied in routine clinical practice to measure side effects accurately.67

Clinician, patient and treatment factors

Some patients will not report symptoms because they are too embarrassed or feel nothing can be done.10 Clinicians may not understand the significance of patients' symptoms or their relationship to previous cancer treatments or simply ignore them.61 Health professionals seeing these patients need to develop strategies to identify proactively unexpected symptoms potentially amenable to treatment.

Different problems and rates of late effects occur depending on specific treatments and how they are combined. Table 2 shows the rate and nature of problems for a wide variety of cancers and table 3 shows how different treatments for the same cancer can have different toxicity profiles.

The non-specific nature of these post-treatment symptoms, which often occur in combination, requires a systematic approach to unravel the associated, sometimes complex, clinical causes. In patients attending a specialist clinic for late GI effects after cancer therapy, more than half were found to have more than one cause for their symptoms.50 For each specific symptom there are a number of potential diagnoses. Using a systematic algorithmic approach, there are standard tests which may elucidate the diagnosis or diagnoses causing each specific symptom.3 50 51 94 95 For each diagnosis made, a number of possible established and experimental therapies are available. Anecdotal evidence suggests that such a systematic approach can improve symptoms by an average of 70–90%. The efficacy of such an algorithmic approach in patients with GI symptoms after pelvic radiotherapy is currently being tested in a large, almost completed, randomised clinical trial (ORBIT study, ISRCTN 22890916).

Managing symptoms

A systematic approach to managing symptoms will only function usefully if a clinician can clearly elicit an accurate history from the symptomatic patient who has often undergone complex multimodal treatments. Many patients do not understand precisely the details of previous cancer treatments, which sometimes have gone on for years and may still be continuing, and are frequently frightened that any persistent symptom is a manifestation of recurrent cancer. Contacting the oncologists and surgeons at the cancer centre for precise details of previous treatments frequently changes management.

History

In addition to a detailed history relevant to the GI tract, it is important to remember that, especially after previous pelvic radiotherapy, patients will also often have other urinary, sexual, psychological and occasionally neurological symptoms which influence or are influenced by their GI problems. Bone disease (eg, pelvic insufficiency fractures) and lymphoedema are also common and may also have not been previously detected or dealt with and may be the cause for some or all of their problems.90 92 98 99

Dietary habits—excess fibre (from ‘healthy eating’), inadequate fibre, alcohol excess or unhelpful nutritional supplements (eg, excess selenium causing nausea, diarrhoea and halitosis)—may contribute to or be the sole cause of chronic GI symptoms surprisingly commonly. Consultations which include input from the patient's partner often improve the quality of the history obtained.

Examination, investigations and treatment

An appropriate physical examination is required. Basic initial investigations should include haematological and biochemical profiles, inflammatory and tumour markers. In addition, clinical experience and limited published data suggest that a number of other tests are particularly worthwhile in patients who are symptomatic after cancer treatments (box 1).50

Specific investigations should be tailored for the principal symptoms and should reflect an understanding of the potential aetiologies. For example, there are at least 13 different causes for diarrhoea after pelvic radiotherapy, most of which require different treatments, and five different causes for new-onset steatorrhoea (table 5).

Faecal incontinence affects up to 50% of patients after rectal cancer and one in five patients after pelvic radiotherapy. Few patients are referred for specialist evaluation, let alone support by incontinence services. Most commonly, evaluation is offered by coloproctologists. However, in those who have had pelvic radiotherapy or chemotherapy, faecal incontinence is often at least partly due to small bowel causes leading to intestinal hurry (especially small bowel bacterial overgrowth and bile acid malabsorption). Appropriate investigations are required when loose stool or erratic bowel function is present, and standard therapeutic approaches to faecal incontinence aimed mainly at local anorectal causes have proved to be ineffective.61

Rectal bleeding from radiation-induced telangiectasia after pelvic radiotherapy

The dose of radiotherapy delivered to the anterior rectal wall determines the risk of bleeding from telangiectasia.100 Bleeding occurs in 50% of patients after pelvic radiotherapy but impairs quality of life requiring intervention in fewer than 6%. Telangiectases often heal spontaneously over 5–10 years. Patients with any rectal bleeding should be offered at least flexible sigmoidoscopy because of the high prevalence of unexpected pathology. All currently available interventions (endoscopic, surgical and hyperbaric oxygen therapy) for radiotherapy-induced bleeding are not risk-free. The only four treatments with any evidence of benefit in randomised trials (of very variable quality) are sucralfate enemas,101 4 weeks of treatment with metronidazole,102 vitamin A103 and hyperbaric oxygen therapy (figure 1A,B).42

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

(A) Painful rectal ulceration following argon beam ablation for radiation-induced bleeding after treatment for prostate cancer. (B) Almost complete resolution of ulceration after 40 sessions of hyperbaric oxygen therapy.

Many gastroenterologists consider that argon plasma coagulation (APC) is the treatment of choice. However, it should be used with considerable caution in this patient group. It is not widely appreciated that the published literature can be interpreted as showing that the serious complication rate for APC when used for radiation proctopathy is potentially as high as 26%.9 51 Some of these disastrous complications—such as explosions following use of APC in inadequately prepared bowel—are preventable. Others, such as the occurrence of deep ulceration,104–106 fistulation,107 stricture formation,104 108 109 bleeding,105 106 110 perforation105 and severe sometimes chronic pain,110–112 reflect the risk of any thermal therapy in chronically ischaemic tissues. It may be that restricting argon flow rates and wattage and very precise and brief application of the argon catheter would reduce complication rates,113 but this has not been proved. Anecdotal evidence suggests that, where bleeding is heavy, APC frequently fails. In specialist centres, serious complications of previous APC treatment in this patient group continue to be seen regularly.

The other treatment commonly prescribed is regular application of corticosteroid enemas. The evidence from randomised trials suggests minimal or no benefit from their use.114 This is unsurprising as the chronic pathophysiology of radiation-induced damage is largely ischaemic and not inflammatory.24 If topical treatment is used, sucralfate enemas (box 2) are clearly more effective than corticosteroid enemas.114

All the options for treatment of radiation-induced telangiectasia are listed in table 6. Evidence for long-term outcomes from any of the treatments is very scanty.

In the absence of any comparative studies of the various major treatment modalities of radiation-induced rectal bleeding, one clinical approach for patients with radiation-induced rectal bleeding which reduces risk to a minimum is as follows:

• Step 1: Investigate with flexible endoscopy to determine the cause of the bleeding.

• Step 2: Optimise bowel function and stool consistency which may reduce the amount of bleeding.

• Step 3: If bleeding is not affecting quality of life (eg, staining clothes, causing anaemia, interfering with daily activities), reassure and do nothing further.

• Step 4: If bleeding affects quality of life, stop/reduce anticoagulants if possible and, if very severe, start sucralfate enemas (box 2),

• Step 5: Discuss definitive treatment to ablate the telangiectasia with the patient; current options include:

1. hyperbaric oxygen therapy (advantages: supported by RCT evidence, may reverse progressive changes caused by radiotherapy, may improve other symptoms such as urinary problems; disadvantages: time-consuming (8 weeks of daily treatment), expensive and patients may need to travel long distances to their nearest unit);

2. argon plasma coagulation (advantages: easily available and simple; disadvantages: significant risk and unproven efficacy in heavy bleeding);

3. formalin therapy (advantages: simple to perform; disadvantages: long-term outcomes poorly known, small risks of serum sickness, severe proctitis or chemical burn to the skin if there is spillage).

Dysphagia/retching/nausea after cancer therapies including upper GI surgery

This is a common cause for referral. A suggested approach is given in table 7. Clinical experience suggests that gastric bile reflux and small bowel bacterial overgrowth are commonly forgotten causes of nausea in patients during and after cancer therapy. Strictures are often found on endoscopy and an approach to stricture management in the upper GI tract is shown in table 8. If nausea and vomiting persist after metabolic causes have been excluded, endoscopy with or without radiology have revealed no cause and trials of routine therapy have not helped, a brain scan should be considered.

Dumping syndrome

When gastric emptying is rapid it leads to ‘dumping syndrome’, characterised by GI and vasomotor symptoms that occur after meals. This can be soon after meals or delayed for up to several hours. The physiological causes are complex, but include the high osmolarity of small bowel contents and reactive hypoglycaemia. The diagnosis of early dumping is usually made on clinical grounds, although rarely gastric scintigraphy is helpful. Late dumping can be diagnosed by measuring blood glucose when patients are symptomatic. If the presentation is atypical, the rare possibility of an insulinoma should be considered. Initial management should be dietary advice to reduce the volume and osmolarity of food presented to the small intestine and the avoidance of fluids taken with meals. Loperamide, guar gum or pectin to slow gastric emptying may be helpful. For late dumping, acarbose may sometimes help. Octreotide or lanreotide are helpful in the short term. Studies have shown mixed results when these drugs were evaluated longer term. Surgical revision of the roux-en-Y anastomosis is a complex procedure but can be effective in selected cases.