As outlined by IOM standards, all financial relationships of expert panel members that might result in actual, potential, or perceived conflicts of interest were individually reported. Disclosures were made prior to the onset of manuscript development and updated on an annual basis. In addition, panel members were asked to articulate any actual or potential conflicts at all key decision points during guideline development, so that participants would understand all possible influences, biases, and/or the diversity of perspectives on the panel. Although some degree of relationships with outside interests are to be expected among experts, panel candidates with significant financial connections that may compromise their ability to fairly weigh evidence (either actual or perceived) were not eligible to participate in guideline development.

Recognizing that guideline panel members are among the leading experts on the subject matter under consideration and guideline recommendations should have the benefit of their expertize, any identified potential conflicts of interests were managed as outlined in SITC’s disclosure and conflict of interest resolution policies. As noted in these policies, panel members disclosing a real or perceived potential conflict of interest may be permitted to participate in consideration and decision-making of a matter related to that conflict, but only if deemed appropriate after discussion and agreement by the expert panel.

The financial support for the development of this guideline was provided solely by SITC. No commercial funding was received.

Panel recommendations are based on literature evidence, where possible, and clinical experience, where appropriate.11 Consensus for the recommendations herein was generated by open communication and scientific debate in small-group and whole-group settings, surveying and responses to clinical questionnaires, as well as formal voting in consensus meetings.

For transparency, a draft of this clinical practice guideline was made publicly available for comment during the development process and prior to publication. All comments were evaluated and considered for inclusion into the final manuscript according to the IOM standard.

The evidence-based and consensus-based recommendations of the panel were refined throughout the development process in order to obtain the highest possible agreement among the experts, however, the minimum threshold was defined as 75% approval among the voting members. Evidence supporting panel recommendations was graded according to the Oxford Centre for Evidence-Based Medicine (OCEBM) Levels of Evidence Working Group ‘The Oxford Levels of Evidence 2’ (2016 version). A summary of the OCEBM grading scale is mentioned in table 1. The level of evidence (LE) for a given recommendation is expressed in parentheses following the recommendation (eg, LE: 1). Recommendations without an associated LE were based on expert consensus.

The following recommendations are intended to form a generalized framework for the management of irAEs resulting from ICI treatment, including direction for at what grade of toxicity to hold and/or permanently discontinue therapy as well as optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents. Exceptions to these recommendations or deviations from these general management principles may exist for certain toxicities. These exceptions will be noted in the individual toxicity sections of this manuscript.

Nausea (sometimes with accompanying vomiting) is a relatively common ICI-associated AE, and occurs in 12% of patients being treated with PD-(L)1 inhibitors, 19% of patients being treated with CTLA-4 inhibitors, and 25% of patients being treated with combination CTLA-4 and PD-(L)1 inhibitors. Despite the frequent occurrence of low-grade nausea, events of grade ≥3 are rare and occur in 0%–2% of patients.4 While nausea and vomiting are not typically serious alone, they may be symptomatic of other conditions affecting the upper GI tract, which require intervention, including infection, GI metastases, endocrinopathy, or other irAEs such as gastritis or colitis. A number of effective options exist for the treatment of nausea, and the most appropriate agent depends on assessment of the individual patient.50 51

In addition to nausea, other irAEs of the upper GI tract have been observed, although they are rare. The irAEs that have been observed include gastritis,52 duodenitis,53 and esophagitis.54 55 In the largest case series on upper tract irAEs, gastric involvement was more common than duodenal involvement by both endoscopy and histology.56 Information on the management of these irAEs is sparse, but management techniques often resemble those used for diarrhea and colitis.

Diarrhea is another common irAE. Diarrhea has been reported in approximately 44% (10% grade ≥3) of patients treated with combination CTLA-4 inhibitors and PD-(L)1 inhibitors, 36% (8% grade ≥3) of those treated with CTLA-4 inhibitors, and 11% of those treated with PD-(L)1 inhibitors (1% grade ≥3).4 57 Diarrhea merits close monitoring, since severe diarrhea alone is clinically concerning, and the presence of diarrhea may indicate colitis, another serious and potentially life-threatening irAE.

Colitis is inflammation of the colon, with diarrhea being a common symptom. The diagnosis of colitis may be inferred by symptoms, and alarm symptoms include pain and hematochezia. Colitis occurs in 16% (11% grade ≥3) of patients treated with combination ICIs, 8% (5% grade ≥3) of those treated with CTLA-4 inhibitors and 1% (1% grade ≥3) of those treated with PD-(L)1 inhibitors.4 58 A study of 182 patients with ICI-induced colitis showed that positive fecal lactoferrin and fecal calprotectin >150 μg/g of stool were both predictive of inflammation, as detected by endoscopic and histological findings.59 Fecal lactoferrin and calprotectin are more well-studied in the context of inflammatory bowel disease than in ICI-induced colitis. While data from ICI-induced colitis does correlate with endoscopic findings, this evidence is not currently strong enough to justify the use of these tests exclusively (ie, excluding endoscopy if fecal lactoferrin and calprotectin tests are negative). The incidence of immune-related GI AEs is higher in patients treated with combination anti-CTLA-4/anti-PD-(L)1 regimens compared with ICI monotherapy, but the rates of colitis do not vary between tumor types for patients receiving ICIs targeting the PD-1/PD-L1 axis.58

Colitis can lead to a number of complications, including bowel perforation, ischemia, necrosis, bleeding, and toxic megacolon. The median time until the onset of diarrhea or colitis is 5–10 weeks.29 60 Upon treatment of patients with corticosteroids, symptoms of diarrhea and colitis may be expected to improve or resolve in 1–2 weeks.61 62 In a retrospective study, 34% of patients who were re-challenged experienced recurrence of colitis, and risk factors for recurrence included initial treatment with anti-PD-(L)1 inhibitors, re-challenge with anti-CTLA-4 inhibitors, higher grade of colitis, requirement for immunosuppressive therapy during initial colitis, and longer duration of initial colitis.63 Patients who do not experience an improvement of their symptoms in 3–5 days should be considered steroid-refractory and may benefit from treatment with a tumor necrosis factor (TNF)-α antagonist antibody, such as infliximab or vedolizumab, which block α₄β₇ integrin, resulting in gut-selective anti-inflammatory activity.64–66 Notably, evidence is emerging that early administration of infliximab or vedolizumab (<10 days after symptom onset) may be associated with more favorable outcomes in immune-mediated colitis, including fewer hospitalizations, less frequent steroid taper failures, shorter courses of steroids, and shorter duration of symptoms compared with later treatment.67 There are currently no head-to-head studies comparing infliximab with vedolizumab for the management of ICI-induced colitis.

Immune-modulating medications such as infliximab and vedolizumab may be problematic in patients with latent viral or bacterial infections. Infliximab, for example, has received a black box warning from the FDA due to the risk of reactivation in patients with latent tuberculosis (TB) infection. However, evidence suggests that patients with HIV may safely receive infliximab,68 and that patients with TB, HIV, and hepatitis B virus (HBV) may safely receive vedolizumab.69–71

Colitis frequently merits examination by methods such as colonoscopy, flexible sigmoidoscopy, and/or computed tomography (CT) scan. The diagnosis of immune-related colitis (without diarrhea) is typically based on histology. In cases where factors such as suspected bowel perforation or toxic megacolon exclude performing a colonoscopy, a CT scan is an effective, non-invasive alternative.72 ICI-induced colitis may be inflammatory and possibly ulcerative, and may appear in multiple locations and patterns in the colon. Histological analysis of biopsies may also reveal patterns of immune infiltration associated with chronic, acute, or lymphocytic colitis.73 A prospective study of 37 patients with ICI-induced colitis found that patients treated with anti-CTLA-4 ICIs exhibited enrichment of CD4⁺ T cells, while those treated with anti-PD-(L)1 ICIs exhibited enrichment of CD8⁺ T cells.74

A number of risk factors have been identified that may predict the development of diarrhea and colitis. The gut microbiome of the patient may influence colitis development, with patients enriched in the phylum Bacteroidetes less likely to develop ICI-associated colitis and patients enriched in bacteria of the phylum Firmicutes more likely to develop colitis (although Firmicutes was also associated with enhanced overall survival (OS) and progression-free survival (PFS)).75 76 Prior treatment with non-steroidal anti-inflammatory drugs (NSAIDs) also increases the risk of colitis.77 Patients with an existing history of GI autoimmune disorders, particularly inflammatory bowel disease, exhibit flare-ups in a substantial minority of cases as a result of ICI therapy (although they may develop other irAEs).78–82

Although rare, new-onset celiac disease has been reported after ICI therapy. The initial presentation of ICI-associated celiac disease shares several features with duodenitis, with the most common presentation being abdominal pain, vitamin deficiencies, dermatitis herpitiformis, transaminase elevations, and constitutional symptoms. Diagnosis was established by the presence of tissue transglutaminase antibodies (tTG IgA). Among the eight patients who presented with new-onset celiac disease after ICI, improvement in signs and symptoms (including down-trending tTG IgA) was observed after a gluten-free diet was implemented as the sole intervention.83

In the context of ICI therapy, hepatitis is often asymptomatic and typically manifests as an elevation in alanine transaminase (ALT) and/or aspartate transaminase (AST) serum levels.84 Of note, other sources of elevated ALT/AST include muscle, so some centers add on CK measurement to make sure myositis/myocarditis is not being missed if values are elevated. Hepatic toxicities typically manifest 1–15 weeks after treatment, although as with any irAE they may occur after delays of months or years.29 60 Hepatitis (defined as ALT/AST elevation) occurs in 5% (1% grade ≥3) of patients treated with PD-(L)1 inhibitors, 5% (2% grade ≥3) of patients treated with CTLA-4 inhibitors, and in 19% (9% grade ≥3) of patients treated with combination ICIs.4 Infliximab should not be used in patients with liver injury, given the risk for hepatotoxicity.85 Owing to their use in non-ICI-mediated autoimmune hepatitis, immunomodulatory agents such as tacrolimus and mycophenolate mofetil have been suggested as treatments for steroid-refractory ICI-mediated hepatitis.86 Biopsy can be considered to initially diagnose steroid-refractory ICI-induced hepatitis, or to help identify the cause of steroid failure.

Cholecystitis and cholangitis are forms of hepatobiliary toxicity that are rarely associated with ICI therapy. A number of case reports have described isolated cases of each disease after ICI therapy.87 Due to a low number of total cases, however, it is difficult to estimate the incidence and causal relationship with immunotherapy, if any, of these irAEs. One case series that included >4000 patients estimated the incidence of suspected cholecystitis associated with any ICI as <1%88 and another retrospective analysis of medical records of 91 patients treated with nivolumab reported a cholangitis incidence of 3% (3 cases).89 Emerging evidence is suggesting that biliary complications may represent a distinct presentation associated with anti-PD-(L)1 therapy and that cholangiopathic irAEs may be more prone to steroid-refractoriness.90 91 Case reports have emerged of the successful use of tocilizumab to manage steroid-refractory biliary irAEs,92 93 however, due to relatively sparse data further studies are needed to determine true incidence rates and optimal interventions.

ICI therapy may cause elevated levels of lipase or amylase. However, diagnosis of acute pancreatitis as an ICI-related irAE is rare.94 The clinical diagnosis of pancreatitis requires at least two of three features to be present: elevated lipase to more than 3 times upper limit of normal (ULN), epigastric pain consistent with pancreatitis, and characteristic radiographic imaging.95 Because patients with elevated amylase or lipase are often otherwise asymptomatic for pancreatitis, few cases of ICI-related acute pancreatitis have been documented.94 Additionally, no association has been demonstrated between the degree of lipase elevation and the severity or prognosis of pancreatitis.

In a retrospective analysis of 2,279 patients monitored for lipase elevation, patients receiving anti-PD-(L)1, anti-CTLA-4, or combination ICI therapies exhibited grade ≥3 lipase elevation (ie, >2–5 times ULN with symptoms or >5 times ULN without symptoms) at rates of 4%, 2%, and 8%, respectively. Of the patients with grade ≥3 lipase elevation, 61% exhibited no other symptoms of pancreatitis, while 39% exhibited at least one other typical symptom of acute pancreatitis. The median time from ICI therapy initiation to peak measured lipase elevation was 146 days in patients treated with anti-PD-(L)1 therapy, 69 days in patients treated with anti-CTLA-4 therapy, and 110 days for patients treated with combination ICI therapy (p=0.03). In these cases, symptoms typically resolved to grade ≤1 within roughly 50 days, regardless of steroid usage. Risk factors for adverse outcomes resulting from acute pancreatitis (eg, chronic pancreatitis, diabetes) included longer duration of ICI therapy, history of smoking, and history of hyperlipidemia. The administration of intravenous (IV) fluids during acute pancreatitis was associated with better long-term outcomes, but for patients with grade 3 or 4 acute pancreatitis, the use of corticosteroids does not appear to improve outcomes and management should include consultation with a GI specialist.96

The following recommendations are intended to be used within the framework of toxicity management, including direction for what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any GI-specific exceptions or additional considerations to the general toxicity management principles are noted in the recommendations below.

The following recommendations are intended to be used within the framework of toxicity management discussed in the General panel recommendations section, including direction for what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE. Any fatigue-specific exceptions or additional considerations to the general toxicity management principles are noted in the recommendations below.

Patients treated with ICIs commonly develop rashes, although these rashes do not frequently rise to grade ≥3 severity. Patients develop rashes at a rate of 23% (1% grade ≥3), 10% (<1% grade ≥3), and 41% (5% grade ≥3) in response to therapy with anti-CTLA-4, anti-PD-(L)1, and combination ICI therapies, respectively.4 A variety of rashes have been recorded, including maculopapular, eczema or atopic dermatitis, lichenoid rash, blistering disorders, and pruritus or acneiform; all of these subtypes are often combined as “rashes” in reports of irAE incidence. Rash typically presents early in treatment, with onset from 2–5 weeks after first ICI treatment.134 Rash is often accompanied by pruritus.

Pruritus is a very common irAE, and often occurs in conjunction with rashes or other skin toxicities but may also present with no obvious skin lesions in 50% of cases. Treatment with anti-PD-(L)1, anti-CTLA-4, and combination ICIs is associated with pruritus rates of 15% (0% grade ≥3), 25% (1% grade ≥3) and 34% (2% grade ≥3), respectively.4 Management may include topical corticosteroids, oral antipruritics, and systemic immunomodulators. One retrospective analysis that included a total of 285 patients treated with ICIs across three centers and encompassing 427 skin irAEs found that most cases of pruritus (n=17) showed moderate to substantial improvement in symptoms after the administration of γ-aminobutyric acid (GABA) analogs pregabalin and gabapentin.135

Vitiligo frequently presents concurrently with other irAEs, and especially with other skin irAEs. Vitiligo has been documented primarily in patients with melanoma,136 137 and appears to be associated with response to immunotherapy and survival in patients with melanoma.138 139 However, vitiligo in patients with other types of cancer has been reported, and it is possible that these cases are under-reported due to a lack of systematic skin evaluation outside of patients with melanoma.140–142 A systematic review found that patients treated with ipilimumab (anti-CTLA-4) developed vitiligo at a rate of 4%, those treated with nivolumab (anti-PD-1) at 9%, and those treated with pembrolizumab (anti-PD-1) at 6%.143

More uncommon, potentially severe or life-threatening cutaneous irAEs have been observed in patients treated with ICIs. These include pemphigus, pemphigoid, lichenoid rash, and SJS/TEN.144–146 Diagnosing and distinguishing between these irAEs may require the use of the salt-split skin technique on a skin biopsy sample, which uses skin separated between the epidermis and dermis to enable direct immunofluorescence analysis.147 Pemphigoid appears to be largely associated with anti-PD-(L)1 ICIs,136 148 149 and is less common with anti-CTLA-4 ICIs (primarily appearing in a small number of case reports).150 151 Pemphigoid occurs in an estimated 1% of patients receiving anti-PD-(L)1 therapy.148

SJS/TEN is rare and can be life-threatening, and SJS/TEN has been reported in patients treated with anti-PD-1 and anti-CTLA-4 inhibitors.131 152–154 SJS/TEN is accompanied by mucosal involvement (ocular, oral, or anogenital). SJS/TEN may be rapid and acute, or may appear progressively from a less severe skin toxicity, such as lichenoid eruptions that fail to respond to typical therapies.

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any skin-specific exceptions or additional considerations are noted in the recommendations below.

Arthralgia in numerous sites can occur during ICI therapy. In addition to being a toxicity induced by ICI therapy, arthralgia is also a known consequence associated with some cancers, chemotherapeutics, and radiation.160–163 This complicates the reporting and attribution of arthralgia, since it may be difficult to distinguish the source of arthralgia in the context of ICI cancer therapy. A systematic review reported arthralgia in 1%–43% of patients enrolled in ICI clinical trials.164 In some cases, arthralgia may be a symptom of a more serious inflammatory AE such as inflammatory arthritis.

Similar to arthralgia, it is difficult to accurately report on the incidence of ICI-induced myalgia, since myalgia can also occur due to cancer. Myalgia occurs in 2%–21% of patients participating in ICI clinical trials.164 Myalgias can be an isolated AE, or can be associated with inflammatory myositis.165 166

Inflammatory arthritis is an ICI-associated AE. Patients with ICI-associated arthritis have clinically heterogeneous presentations; some patients have symptoms that are similar to rheumatoid arthritis, others have psoriatic arthritis-like symptoms, and in some rare instances, there are symptoms of a reactive arthritis.164 167–170 In addition to arthralgia, arthritis may be accompanied by symptoms such as joint stiffness and swelling and can affect patients’ ability to perform activities of daily living. Synovitis, tenosynovitis, and/or enthesitis can be detected on physical examination. Patients are mostly seronegative for rheumatoid factor and anti-cyclic citrullinated peptide (anti-CCP) antibodies, but cases of seropositivity have been reported.169 171 172 In a systematic review of ICI-related inflammatory arthritis, 1%–7% of clinical trial participants developed arthritis.164

Symptoms of ICI-induced polymyalgia rheumatica (PMR) include joint pain and stiffness (primarily in the shoulders and hips), and may rarely be accompanied by giant-cell arteritis (GCA), in which case visual symptoms must be assessed.173 174 Assessment of the incidence of PMR is difficult because the majority of studies of musculoskeletal irAEs are observational and retrospective. The median time to onset of PMR is 12 weeks, and the majority of cases respond well to corticosteroids.175 Patients with corticosteroid-refractory PMR may derive benefit from non-steroidal agents such as methotrexate, hydroxychloroquine, or tocilizumab.

Dry mouth (or xerostomia) may occur as a separate irAE, or it may be a symptom of another irAE such as ICI-induced sicca syndrome.176 Dry mouth may also be linked to infections such as oral candidiasis—this is of particular concern in patients treated with ICIs, since oral corticosteroid use (commonly used to combat some ICI toxicities) is a risk factor for oral candidiasis.29 177 178 Dry mouth may be the cause of complications including infection and dental pathologies, including the loss of teeth, in severe cases of sicca syndrome.178 179 For this reason, it is important to account for the possibility of infection during diagnosis and monitoring. A systematic review determined that many clinical trials did not report the overall incidence of dry mouth, but that incidence varies from 3%–24% depending on the study.174 For initial management of dry mouth or painful sores, oral rinses with doxepin mouthwash or diphenhydramine-lidocaine-antacid mouthwash (sometimes called ‘magic mouthwash’) have been shown to reduce radiotherapy-related mucositis pain,180 181 but data are lacking on their efficacy in ICI-treated patients. Sialogogue therapy, cevimeline, pilocarpine, or other systemic acetylcholinergic agents have been anecdotally reported as helpful when symptoms are persistent and bothersome and refractory to topical mouth rinses.

ICI-induced sicca syndrome may feature severe symptoms, including dry eyes and dry mouth. Sicca accompanied by true Sjögren syndrome (typically with the presence of ANAs, anti-Ro, and/or anti-La antibodies) is rare, occurring at a rate of <1% in a registry study of grade ≥2 irAEs.167 The median time to onset of dry mouth, indicative of sicca syndrome, is 70 days. Patients who develop sicca syndrome often do not experience full resolution of their symptoms, and they may require long-term care for salivary hypofunction as well as being at risk for loss of teeth.176

Vasculitis resulting from ICI treatment is a rare irAE and has been reported involving large vessels, medium vessels, small vessels, and the central nervous system.182 183 GCA, a type of large vessel vasculitis, has been seen with PD-(L)1 and CTLA-4 blockade and may be associated with PMR symptoms. No deaths have been attributed to ICI-induced vasculitis, and symptoms in each case resolved following withholding of ICIs and/or corticosteroid therapy. Induction with rituximab or cyclophosphamide have also been used in addition to steroids. Vasculitis typically occurs at a median of 3 months following initial ICI treatment.182 In those patients with suspected GCA, a low threshold for temporal artery biopsy in consultation with a rheumatologist and urgent corticosteroids may be warranted due to risk of visual loss.

Myositis is a rare, potentially serious irAE, occurring in 1% of patients treated with anti-PD(L)1 ICIs and <1% of patients treated with anti-CTLA-4 ICIs.46 Little systematic data exist regarding the occurrence of myositis during treatment with other ICIs. Myositis is often associated with other serious muscular and neurological dysfunction, including myocarditis (9% of patients with ICI-associated myositis) and myasthenia gravis (9% of patients with ICI-associated myositis).46 Fatalities in patients with myositis may arise from associated irAEs such as myocarditis, or may occur directly as a result of myositis, particularly if there is diaphragmatic or respiratory muscle involvement. Symptoms of myositis include restricted eye movement, problems with speaking or swallowing, muscle weakness in the limbs, and myalgia. Some patients are asymptomatic, but exhibit elevated levels of CK. Conversely, some symptomatic patients present with normal CK levels. Myositis may require continued monitoring and/or follow-up treatment—in one analysis, 50% of myositis cases were ongoing or had resulted in sequelae at the end of the 5-year observation period.48

The following recommendations are intended to be used within the framework of toxicity management, including direction for what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. For suspected myositis, myocarditis, and myasthenia gravis, there is a possibility of overlapping symptoms, and therefore patients should be evaluated with a shared set of diagnostics, as described in more detail in the General panel recommendations section. Any additional musculoskeletal-specific exceptions or additional considerations are noted in the recommendations below.

Dry eyes may be an AE in their own right or may be a symptom of another AE such as sicca syndrome.168 The dry eye syndrome associated with ICI treatment may be severe enough to cause corneal perforation.200 The incidence of dry eyes in prospective trials ranges from 1%–24%.201

Uveitis typically presents with symptoms such as eye redness, pain, photophobia, floaters, and blurred vision.196 202 Uveitis induced by ICI therapy may be anterior, posterior, or panuveitis.203–209 Uveitis occurs at a rate of <1%–6% in clinical trials, based on a systematic review of the literature.201 It is important to note that symptoms of uveitis may not reflect the potential severity of the condition, and that uveitis can also occur as a result of infectious causes or metastasis to the eye. Therefore, a careful assessment is necessary to exclude other causes that require different treatment and to ensure proper management is initiated to prevent vision loss.210–212 Metastasis to the eye is especially concerning, since the eye is an immune-privileged site and could conceivably be difficult to treat with immunotherapies.212 Several steroid-sparing agents for non-infectious uveitis have been investigated outside of the context of ICI therapy, including methotrexate, mycophenolate mofetil (which did not demonstrate superiority to methotrexate),213 and secukinumab.214

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any opthamological-specific exceptions or additional considerations are noted in the recommendations below.

Hypothyroidism and hyperthyroidism are frequently asymptomatic or exhibit ambiguous symptoms, necessitating routine monitoring of parameters such as TSH and total T3/fT4 levels.222–224 Hypothyroidism is the more common of the two toxicities, occurring in about 8% of patients receiving anti-PD-(L)1 therapy, 3% of patients receiving anti-CTLA-4 therapy, and 15% of patients receiving combination ICI therapy. Grade ≥3 hypothyroidism is rare, occurring in roughly 0%–2% of patients receiving combination ICI therapies.4 The standard of care for the treatment of hypothyroidism is levothyroxine.225

Hyperthyroidism occurs less frequently, in 5% of patients treated with anti-PD-(L)1 inhibitors and 4% of patients treated with anti-CTLA-4 inhibitors.4 Rarely, ICI therapy may lead to Graves’ disease.226 Symptoms of elevated thyroid hormone may also appear transiently and evolve into hypothyroidism215 as a result of patients experiencing thyrotoxicosis during the course of thyroiditis, due to the destruction of thyroid follicles and necrosis.227 Hypothyroidism frequently occurs following this transient hyperthyroidism, as a sequela of ongoing thyroiditis—roughly 90% of patients who develop thyrotoxicosis do not recover full thyroid function, requiring long-term levothyroxine replacement. The median time to thyrotoxicosis is 5 weeks, and the median time to hypothyroidism is 10 weeks.215

Hypophysitis is an uncommon but important irAE, and is often accompanied by symptoms such as fatigue, nausea, vomiting, weakness, headache, and gonadotrophic deficiency (including loss of libido or erectile dysfunction).228 229 ICI-induced hypophysitis is most frequently manifested as secondary adrenal insufficiency due to ACTH deficiency, and less commonly due to TSH, follicle-stimulating hormone (FSH), and luteinizing hormone (LH) deficiency.216 230 Because hypophysitis can be induced through both ICI therapy and through the appearance of pituitary metastases, it is vital that a brain scan (preferably a pituitary-protocol MRI) be conducted to distinguish hypophysitis from alternative causes of headache, including brain metastases.231 Typical findings on MRI indicative of ICI-associated hypophysitis include geographic hypoenhancing lesions in the anterior lobe of the pituitary gland.232 It is further important to note that cortisol and ACTH test results may be inaccurate if patients are receiving steroids at baseline, for example, in patients with lung cancer treated with concurrent chemo and checkpoint inhibitors accompanied by dexamethasone premedication.

Hypophysitis occurs in 1% of patients treated with anti-PD-(L)1 ICIs and in 4% of patients treated with anti-CTLA-4 ICIs (0% and 2% grade ≥3, respectively).4 Patients receiving ICI combination therapy develop hypophysitis at a rate of 9%–11%.233 In an analysis of 689 patients who developed ICI-related hypophysitis, the median time between ICI treatment initiation and diagnosis of hypophysitis was 76 days.234

Rarely, in addition to secondary adrenal insufficiency caused by disruption of the pituitary gland, irAEs have been noted in which the adrenal glands are directly damaged following ICI therapy, leading to primary adrenal insufficiency.235 As with secondary adrenal insufficiency, primary adrenal insufficiency can lead to life-threatening adrenal crisis due to vasodilatory shock.236 The symptoms of adrenal insufficiency, like other endocrinopathies, can be non-specific and difficult to diagnose without additional testing. Symptoms may include, nausea, loss of appetite, weight loss, fatigue, light-headedness, hypoglycemia, and hypotension.236–238 Primary adrenal insufficiency is rare in ICI-treated patients and would be associated with high serum ACTH levels in conjunction with low morning serum cortisol.101 However, morning cortisol and ACTH levels are not always definitive—an ACTH stimulation test is considered the standard of care assessment for adrenocortical insufficiency.

Adrenal insufficiency is not a common ICI-associated irAE, but it necessitates vigilance and close monitoring due to the possibility of adrenal crisis. Adrenal insufficiency occurs in 1% of patients treated with anti-PD-(L)1 or CTLA-4 therapies, and is estimated to occur at a rate of 5% in patients treated with combination ICIs.4 233 In a review of case reports, the median onset of adrenal insufficiency measured from the first dose of ICI therapy was 10 weeks.235

Type I diabetes mellitus (T1DM) induced by ICI therapy is rare, but frequently serious. The incidence of ICI-induced diabetes is estimated at approximately 1%.239 240 T1DM may develop shortly after the beginning of ICI treatment or as much as 1 year following the start of treatment.241 242 Diabetes occurs at a rate of <1%–2% in patients treated with anti-PD-(L)1 inhibitors.233 Diabetes has occurred in patients treated with anti-CTLA-4 inhibitors, but this is a very rare occurrence.243 Often, the initial presentation is fulminant T1DM and diabetic ketoacidosis (DKA). However, patients may be asymptomatic or present with symptoms such as fatigue, nausea, vomiting, weight loss, polyuria, or polydipsia.241 243 Autoantibodies indicative of T1DM are found in up to 53% of cases, with the majority being anti-glutamic acid decarboxylase (GAD-65) antibodies,244 although insulin autoantibodies have also been reported.240 Almost all patients with T1DM require insulin therapy for management.

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any endocrine-specific exceptions or additional considerations are noted in the recommendations below.

Myasthenia gravis is a significant neurological toxicity with a high potential for patient fatality.46 193 Patients present with fatigable or fluctuating muscle weakness, generally involving proximal muscles (neck and shoulder) more than distal muscles. There may be diaphragmatic weakness resulting in respiratory compromise. Bulbar and ocular muscles are commonly affected in myasthenia gravis resulting in ptosis, extraocular movement abnormalities leading to double vision, facial weakness, and difficulty swallowing. Autoantibodies against the acetylcholine receptor (AChR) or MusK may be present in ICI-related myasthenia gravis, but toxicity can occur independent of positive serology. One retrospective analysis that included 47 total patients with ICI-related myasthenia gravis reported serological positivity rates by anti-AChR antibody and anti-MuSK antibody of 66.7% (30/45) and 5.3% (1/19), respectively.250 Patients with myasthenia gravis may also develop myocarditis and/or myositis as part of a potentially dangerous combination of pathologies.46 251 Patients treated with anti-PD-(L)1 ICIs are at greater risk of developing myasthenia gravis than patients treated with anti-CTLA-4 ICIs.46 The standard of care for myasthenia gravis includes IVIG and plasma exchange (PLEX), and in patients with autoimmune myasthenia gravis also includes high-dose pulse corticosteroids.252–254

ICIs may result in encephalitis at an incidence estimated to be <1%. Patients may present with a wide range of symptoms including altered behavior, confusion, short-term memory impairment, agitation, speech abnormality, and seizures. Treatment with anti-PD-(L)1 monotherapies or with combination ICIs is associated with a higher risk of encephalitis compared with treatment with anti-CTLA-4 ICIs.46 Rarely, patients have been found to have positive autoimmune encephalitis or paraneoplastic neurological syndrome antibodies.255–257

Peripheral neuropathies may be observed in a number of different phenotypic presentations, including painful small fiber sensory type and cranial neuropathies,258 259 as well as sensorimotor presentations more typical of classic immune-mediated phenomena, such as Guillain-Barré syndrome.46 260 Guillain-Barré syndrome is more common in patients treated with anti-CTLA-4 or combination anti-CTLA-4 plus anti-PD-1 ICIs when compared with patients treated with anti-PD-(L)1 ICIs alone.46 Guillain-Barré syndrome, or ascending polyradiculoneuropathy, can develop soon after ICI treatment is started, usually within the first 3 cycles. Patients may develop early lower back or thigh pain followed by ascending weakness, sensory loss, and areflexia as the main symptoms. Facial weakness and extraocular movement impairment as a result of cranial neuropathies may occur. There may also be dysregulation of autonomic nerves. Nerve root enhancement and thickening may be seen on imaging.

Patients may develop other acute neuropathies, including painful sensory neuropathy and isolated cranial mononeuropathies, especially of the facial and abducens nerves.258 Patients with painful neuropathies may require pain management.

Patients may present with headache, neck stiffness, photophobia, low-grade fever, and nausea. Typically, mental status is normal in these patients (in contrast with encephalitis). As with any case of meningitis, the possibility of infectious meningitis must be seriously considered during diagnosis and treatment and leptomeningeal metastasis should also be ruled out. Patients with headache should also have hypophysitis ruled out. Meningitis is associated more frequently with anti-CTLA-4 and anti-CTLA-4 plus anti-PD-(L)1 combination therapies than with anti-PD-(L)1 monotherapies.46

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. For suspected myositis, myocarditis, and myasthenia gravis, there is a possibility of overlapping symptoms, and therefore patients should be evaluated with a shared set of diagnostics, as described in more detail in the General panel recommendations section. Any other neural-specific exceptions or additional considerations are noted in the recommendations below.

Pneumonitis is a relatively common irAE that is associated with lower rates of patient survival.102 272 Symptoms include dyspnea, persistent cough, chest pain, fever, and hypoxia (potentially leading to respiratory failure).273 Patients with pneumonitis may also be asymptomatic, yet show detectable inflammation on CT scan.273 274 Radiological findings with ICI-associated pneumonitis may vary, with distinctive features seen on imaging, including discrete patchy or confluent consolidation with or without air bronchograms and predominantly peripheral or subpleural distribution, ground-glass opacities, centrilobular nodules, bronchioloitis-like appearance with tree-in-bud micronodularity or patterns that do not clearly fit within other classifications.273 Characteristic CT scans showing common presentations of ICI-associated pneumonitis are shown in figure 2.

Pneumonitis is associated with the expansion of inflammatory T cell subsets, in both bronchoalveolar lavage samples and from direct tissue samples of pneumonitis-related lesions.275 276 In published meta-analyses, patients with NSCLC or renal cell carcinoma (RCC) have a higher risk of developing pneumonitis compared with patients with melanoma following treatment with PD-(L)1 inhibitors.277 Recent literature indicates that a history of asthma and/or smoking may predispose patients to developing higher grade pneumonitis with ICI therapy.278 Other risk factors that influence the incidence of ICI-associated pneumonitis include prior curative-intent radiotherapy and squamous tumor histology.279 280 Pneumonitis that does not improve with corticosteroid treatment may also be more common in patients who are former or active smokers, or those who have underlying lung conditions.273

Pneumonitis occurs in 4% of patients receiving PD-(L)1 inhibitors (1% grade ≥3) and 1% of patients receiving anti-CTLA-4 therapy (1% grade ≥3).4 The rate of pneumonitis in patients treated with combination ICIs is significantly higher than the rate for patients treated with ICI monotherapies, at 7% (2% grade ≥3).277 A systematic review also determined that PD-1 inhibitors cause pneumonitis at a higher rate than PD-L1 inhibitors in patients with NSCLC.281 Notably, the incidence of pneumonitis recorded in clinical trials may be lower than that observed in routine clinical practice. A retrospective study at a single institution recorded an incidence of 19%, with 12% of patients experiencing pneumonitis of grade ≥3.280 An analysis of ICI-treated patients that developed pneumonitis found that the median onset of pneumonitis is 3 months, and in one clinical trial the median time to resolution of symptoms was 3–4 weeks.273 282

Recurrent pneumonitis following improvement of symptoms has been observed both in patients who were re-challenged with ICI therapy and in patients who were not, demonstrating the need for careful monitoring of all patients with ICI-induced pneumonitis after resolution of symptoms. In a small sample of patients who were re-challenged with ICI therapy following complete clinical resolution of pneumonitis (n=12), 25% developed recurrent pneumonitis that resolved by reinstating the same initial intervention: withholding of ICIs (grade 1) and/or corticosteroid treatment (grade 2).273 In recent studies, pneumonitis has also been found to assume a chronic course in approximately 2% of patients with NSCLC or melanoma treated with ICIs.283 Patients with steroid-refractory ICI-induced pneumonitis may benefit from mycophenolate mofetil or high-dose IVIG (hdIVIG), although this has only been reported in case studies.280 284 Based on its application in other immune-mediated lung diseases, cyclophosphamide might also be expected to provide benefit in steroid-refractory pneumonitis.285 286 However, little data exist that directly tests the use of cyclophosphamide in this treatment context. One study of 120 patients with chronic hypersensitivity pneumonitis (CHP) found that those who were able to tolerate a full course of infusions with cyclophosphamide (6 infusions at 600 mg/m²) appeared to stabilize in their decline in lung function.287 Other approaches that have been reported include anti-cytokine agents. Infliximab (5 mg/kg, IV) has been used to treat patients with ICI-induced pneumonitis, with mixed results from a number of case reports.273 288–290 A study of the IL-6R antagonist antibody tocilizumab (4 mg/kg, IV) for the treatment of nivolumab-associated irAEs also showed potential benefit in the setting of grade ≥3 pneumonitis.291 An ongoing prospective study through ECOG-ACRIN will compare the efficacy of IVIG versus infliximab for steroid-refractory pneumonitis (NCT04438382).

Due to the rarity of ICI-associated sarcoidosis, it is difficult to definitively assess the incidence of this irAE.292 However, it is important that clinicians recognize the possibility of sarcoidosis, since the formation of granulomas associated with this irAE may resemble disease progression or recurrence.292 293 Sarcoidosis may arise in a number of sites within the body, most commonly the lymph nodes (71% of cases), lungs (60% of cases), and skin (55% of cases). Sarcoidosis occurs at a mean of 9 months following the start of ICI treatment, and symptoms resolve in a mean of 4 months.292 Because sarcoidosis is usually asymptomatic and may mimic progressive disease, especially with lymph node involvement, biopsy may be considered in the differential.

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any pulmonary-specific exceptions or additional considerations are noted in the recommendations below.

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any infusion reaction-specific exceptions or additional considerations are noted in the recommendations below.

Acute kidney injury (AKI) describes a condition in which kidney function is severely impacted or lost and may occur via a number of etiologies including prerenal disease, acute tubular necrosis, TIN, autoimmune reactivation of membranous nephropathy, and glomerular diseases.104 304 306–308 AKI can be graded using the CTCAE scale but may also be graded with the Kidney Disease: Improving Global Outcomes (KDIGO) criteria, which defines AKI through the specific criteria of serum Cr levels and urine output, as opposed to the more vague criteria outlined in the CTCAE.309 The KDIGO criteria may enable more detailed classification of kidney disease, since KDIGO stages disease severity using both elevation above baseline and absolute thresholds in serum Cr levels, while the CTCAE only measures based on elevation above baseline.309 310 However, the majority of clinical trials use the CTCAE scale of severity, which may result in underestimation of the rate of AKI. Furthermore, AKI is common in patients receiving ICI therapy, but in most cases, it is not the direct result of ICI toxicity. It is important to differentiate between all-cause AKI (eg, due to hypovolemia or acute tubular necrosis (ATN)) and ICI-induced AKI to ensure appropriate management.311 With these limitations in mind, the recorded incidence of ICI-induced AKI in patients from pooled clinical trials receiving anti-PD-(L)1 ICIs was 2% and the incidence for patients receiving ipilimumab was 2%.304 312 Combination therapy with anti-PD-(L)1 and anti-CTLA-4 ICIs resulted in an AKI incidence of 5%.304 The majority of TIN cases are steroid responsive. Patients with steroid-refractory TIN may benefit from mycophenolate mofetil, based on its efficacy in non-ICI-induced cases of interstitial nephritis.313–315 Other options, including rituximab, have been explored for some glomerular and renal vasculitis diseases.308 316

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any kidney-specific exceptions or additional considerations are noted in the recommendations below.

Toxicities affecting the heart are uncommon with ICI therapy, but important due to the high mortality rate when they overtly develop. In a retrospective study of AEs in patients receiving ICIs, death occurred in 50% of patients who developed myocarditis and in 21% of patients who developed pericardial disease.319 Owing to the low number of events, data on incidence divided by drug and by specific cardiovascular toxicities (eg, myocarditis, pericardial disease) are limited. However, data gathered from an eight-center patient registry of 964 patients treated with ICIs found that the overall incidence of myocarditis was 1% (n=35). In the analysis, nearly all myocarditis cases had elevated troponins (94%) and an abnormal EKG (89%), while left ventricular ejection fraction (LVEF) was normal in 51% of cases. Among the 35 patients with myocarditis, 16 (46%) experienced a major adverse cardiac event, including ventricular arrythmias and complete heart block, cardiogenic shock, cardiac arrest, or cardiovascular death. In this analysis, myocarditis was associated with diabetes mellitus, sleep apnea, and high BMI, and occurred at a median onset of 34 days following the start of ICI therapy.47 The initial diagnosis for suspected cases of ICI-induced myocarditis typically looks for elevated troponin levels and EKG changes, followed by cardiac MRI, and finally the gold standard for myocarditis: endomyocardial biopsy (endovascular). Baseline and subsequently scheduled troponin levels can be obtained, but in asymptomatic patients there has been no evidence that this improves outcome or even provides an early indicator of possible myocarditis. Patients with myocarditis are generally initially treated with high-dose corticosteroids. Early treatment with steroids may be important, as shown in a retrospective analysis including 126 patients with ICI-induced myocarditis from 23 different sites that found administration of corticosteroids within 24 hours of admission led to a lower rate of major adverse cardiac events (7.0%) than treatment between 24 and 72 hours (34.3%) and >72 hours after intake (85.1%; p<0.001).320 Those resistant to corticosteroids may benefit from therapies including ATG, mycophenolate mofetil, abatacept, or alemtuzumab.47 321–323 A retrospective analysis of 60 patients who developed ICI-associated myocarditis found that those requiring second-line immunosuppressives had higher all-cause mortality than those treated with steroids alone (50% vs 21%; p=0.02). Moreover, infliximab use was associated with increased risk of death from cardiovascular causes (odds ratio (OR) 12.0; 95% CI 2.1 to 67.1; p=0.005).324

Attribution of thromboembolic events to ICI therapy may be difficult, since these events may occur as a result of cancer.325 326 Venous thromboembolism (VTE), in fact, is one of the most common causes of mortality in patients with cancer.327 It is estimated that between 4% and 20% of patients with cancer will develop VTE.328 Case reports have recorded thromboembolic events that appear to be temporally associated with ICI therapy.329–331 However, attempts to estimate the overall incidence of thromboembolic events in ICI-treated patients (roughly 8% in a retrospective study by Gutiérrez Sainz et al332) have not distinguished between events induced by ICI therapy and events induced by the malignancy itself (or thromboembolic events induced by other therapies, such as chemotherapy). Regardless of the ultimate etiology of the thrombotic event, oral apixaban has been shown to be non-inferior to subcutaneous dalteparin in a multinational randomized trial,333 and subsequent meta-analyses support direct-acting oral anticoagulants as standard of care for cancer-associated venous embolisms.334

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. For suspected myositis, myocarditis, and myasthenia gravis, there is a possibility of overlapping symptoms, and therefore patients should be evaluated with a shared set of diagnostics, as described in more detail in the General panel recommendations section. Any additional cardiac-specific exceptions or additional considerations are noted in the recommendations below. Additional consideration for the management of cardiac-related AEs may be found in the series of American Heart Association (AHA)/American College of Cardiology (ACC) guidelines.

HA associated with ICI therapy (autoimmune HA (AIHA)) is one of the most common ICI-associated hematological irAEs. AIHA occurs in patients treated with ICIs at a median of 50 days after the initiation of treatment.103 The median time for AIHA to resolve below grade 2 symptoms is 2 weeks.344 Treatment with anti-PD-(L)1 ICIs appears to present a significantly greater risk of AIHA than treatment with anti-CTLA-4 ICIs.345

Thrombocytopenia is a fairly common occurrence in patients with cancer, but the recorded incidence of immune-related thrombocytopenia (immune thrombocytopenia (ITP)) is low.346 Thrombocytopenia can result from cancer itself, chemotherapy, other medications, disseminated intravascular coagulation (DIC), radiation therapy, some infections, and pre-existing autoimmunity (particularly in patients with hematological cancers).347–356 Because patients may already be experiencing thrombocytopenia as a result of cancer or due to a number of oncological treatment regimens, it may be prudent to establish a baseline platelet count and monitor for additional drops below this value when beginning ICI therapy.357 The median time to onset for ITP is 41 days, and the median time for symptoms to resolve to below grade 2 is 4 weeks.103 344 Steroid-refractory ITP has been successfully managed with rituximab.358 359

Hemaphagocytic lymphohistiocytosis/macrophage activation syndrome (HLH/MAS) is a hematological disorder involving inappropriate macrophage activation, with a high potential for lethality. The diagnostic criteria include cytopenias, hyperferritinemia (commonly in thousands of ng/mL), fever, splenomegaly, coagulopathy, LFT abnormalities, and elevation of soluble IL-2 receptor. Examination of bone marrow aspirate by a pathologist can also be important to identify potential HLH.360 361 HLH occurs at a median of 26 days, and is more common in patients treated with anti-CTLA-4 ICIs than anti-PD-(L)1 ICIs.103 The management of HLH/MAS is complex and not ‘one size fits all,’ potentially encompassing cytotoxic chemotherapies such as etoposide, cytokine modulation with agents such as anti-IL-6, and/or corticosteroids, depending on the etiology and patient characteristics.361

Aplastic anemia (AA) is often a result of immune-related bone marrow failure, and manifests as pancytopenia. AA has a low rate of resolution and is frequently difficult to treat.344 ATG has been attempted as a salvage therapy in a patient who developed lethal AA after dual PD-1/CTLA-4 inhibition for metastatic melanoma that did not improve after granulocyte colony-stimulating factor (G-CSF), tranexamic acid, and repeated platelet transfusions,362 but data are lacking on optimal management and options are limited.

Pure red cell aplasia (PRCA) can be associated with either anti-CTLA-4 or anti-PD-(L)1 ICIs but occurs rarely. PRCA symptoms present at a median of 89 days following the start of ICI treatment.363

ICI-related neutropenia typically occurs at a median of 10.5 weeks after the initiation of ICI therapy. As with any form of neutropenia, ICI-related neutropenia carries an increased risk of infectious complications.364

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any hematological-specific exceptions or additional considerations are noted in the recommendations below.

Only one combination of ICIs is FDA-approved, nivolumab with ipilimumab, which has indications for the treatment of unresectable or metastatic melanoma, metastatic NSCLC, advanced RCC, hepatocellular carcinoma (HCC), microsatellite instability high (MSI-H) or mismatch repair deficient (dMMR) metastatic colorectal cancer, and mesothelioma.371 372 The dosing and timing for each ICI in the combination regimen is different across approved disease settings—for example, for HCC, ipilimumab is given at 3 mg/kg with nivolumab at 1 mg/kg every 3 weeks, whereas for metastatic NSCLC, ipilimumab is given at 1 mg/kg every 6 weeks with nivolumab 3 mg/kg every 3 weeks, and other schedules are approved as well. As discussed previously, combination ICI therapy generally results in a higher incidence of all-grade irAEs and of grade ≥3 irAEs. Combination therapy is also associated with higher rates of fatal toxicity—a meta-analysis encompassing >16,000,000 adverse drug reactions in records from 7 academic centers found that among the 613 fatal ICI-associated toxicities reported, the most common causes of death for combination anti-PD-(L)1/anti-CTLA-4 therapy were colitis (n=32; 37%) and myocarditis (n=22; 25%).373

Data are sparse for head-to-head comparisons of toxicities with ICI combination regimens at different dosing regimens in the same disease state. Of note, the phase IIIb/IV CheckMate 511 study, which enrolled patients with advanced melanoma, showed that combination regimens incorporating ipilimumab at 1 mg/kg are associated with fewer severe toxicities, despite comparable response rates. In the study, patients were randomized 1:1 to nivolumab at 1 mg/kg with 3 mg/kg ipilimumab or nivolumab at 3 mg/kg with ipilimumab at 1 mg/kg every 3 weeks for 4 doses. Objective response rates and PFS were not significantly different between the arms, however, the incidence of grade 3–5 TRAEs was significantly higher among the patients being treated with the higher dose of ipilimumab (48% vs 34%; p=0.006).374 Similarly, the CheckMate 040 trial, which randomized patients with HCC 1:1:1 to either nivolumab 1 mg/kg with ipilimumab 3 mg/kg every 3 weeks for 4 doses followed by nivolumab 240 mg every 2 weeks, nivolumab 3 mg/kg with ipilimumab 1 mg/kg every 3 weeks for 4 doses followed by nivolumab 240 mg every 2 weeks, or nivolumab 3 mg/kg every 2 weeks with ipilimumab 1 mg/kg every 6 weeks, found higher incidence of irAEs overall as well as more irAEs requiring discontinuation of treatment in the group receiving high-dose ipilimumab.375

An expanding number of specific combination regimens involving chemotherapy and ICIs have been approved by the FDA, including indications for breast cancer, head and neck squamous cell carcinoma, and lung cancer. Clinical trials involving ICI-chemotherapy combinations have reported toxicity profiles that generally correspond to the additive effects of each agent as monotherapy; combination regimens have not thus far resulted in new AE signals.376–380

ICIs in combination with a variety of small molecule tyrosine kinase inhibitors (TKIs) or anti-angiogenic antibodies have been approved in a variety of disease settings, such as HCC, RCC, NSCLC, and melanoma. The approved combination regimens involving TKIs with ICIs have generally thus far resulted in toxicity profiles similar to those of each agent administered as a monotherapy (eg, rash and diarrhea with TKIs, and hypertension with anti-vascular endothelial growth factor receptor (VEGFR) antibodies), although the combination of pembrolizumab with axitinib appeared to cause a higher incidence of grade ≥3 AEs.381 382 A multitude of ongoing studies are investigating ICIs with other small molecule inhibitors against a variety of targets. Other combinations, such as ipilimumab with vemurafenib,383 durvalumab with osimertinib,384 nivolumab with ademaciclib,385 and ipilimumab with dabrafenib and trametinib,386 have exhibited unexpectedly high rates of toxicity, leading to clinical trial discontinuations. Severe liver toxicity in addition to increased incidence of pneumonitis has been seen with the combination of epidermal growth factor receptor (EGFR) TKIs and anti-PD-(L)1 ICIs,384 387 emphasizing the need for caution in the application of combination immunotherapy regimens. The management of toxicities arising from combination regimens may be complex, and thus consultation with appropriate specialists (eg, cardiology in cases of cardiac AEs with ICI/VEGFR TKI combinations) should be top-of-mind.

The following recommendations are intended to be used within the framework of toxicity management, including direction for at what grade of toxicity to hold and/or permanently cease treatment, considerations for life-threatening toxicities, as well as recommendations on optimal timing and dosing for administration of corticosteroids and/or other immunosuppressive agents by grade of AE, discussed in the General panel recommendations section. Any exceptions specific to combination ICI therapies or additional considerations are noted in the recommendations below.