Following approval of institutional review boards, data were extracted from the medical records of patients at 10 international participating centers.

Patients who had received at least one dose of combination ipilimumab and anti-PD1 between 2015 and February 2020 with a concomitant diagnosis of an autoimmune disorder were included. Qualifying autoimmune disorders included but were not limited to the following: rheumatologic (rheumatoid arthritis (RA), systemic lupus erythematosus, psoriatic arthritis, vasculitis, polymyalgia rheumatica, scleroderma, Sjogren’s syndrome), gastrointestinal (Crohn’s disease, ulcerative colitis, celiac disease), neurologic (Guillain-Barre syndrome (GBS), transverse myelitis, multiple sclerosis, myasthenia gravis, chronic inflammatory demyelinating polyneuropathy), endocrine (Graves’ disease, Hashimoto’s thyroiditis, type 1 diabetes mellitus), dermatologic (psoriasis, eczema, erythema nodosum) and other (sarcoidosis, asthma, idiopathic thrombocytopenic purpura). Autoimmune disorders were diagnosed based on each center’s standard of diagnosis, for most conditions, a history and serological testing confirmed the diagnosis. For patients with IBD and dermatologic conditions, all had a biopsy confirming the diagnosis.

Baseline patient demographics were collected including age, gender, Eastern Cooperative Group Performance Status (ECOG) and prognostic factors including eighth edition of the American Joint Committee on Cancer pathologic stage, presence of brain and liver metastases and serum lactate dehydrogenase level. Severity of the baseline autoimmune disorder was assessed by clinical activity as deemed by the treating clinician, use of baseline immunosuppression and dose, and recent flare of autoimmune disease.

The safety of combination immunotherapy was assessed by worsening of the autoimmune disorder or flare requiring systemic or immunosuppressive therapy or interruption to immunotherapy as well as the incidence and management of conventional irAE. Flare of autoimmune disease was as diagnosed by the patient’s treating oncologist, and where necessary in conjunction with autoimmune disease experts. Severity of irAE and flare was defined by Common Terminology Criteria for Adverse Events criteria.16 The efficacy of combination immunotherapy in this population was measured either by Response Evaluation Criteria in Solid Tumours V.1.117 or by clinical assessment of fluorodeoxyglucose-positron emission tomography (FDG-PET)imaging and assessment of objective response rate, duration of response, progression-free survival (PFS) and overall survival (OS).

Categorical and continuous variables are summarized using percentages and medians. No formal hypothesis testing was performed. OS and PFS were estimated using the Kaplan-Meier method and compared using the log-rank test; all patients were censored at last available follow-up. PFS was defined as time of treatment start to disease progression (as determined by the treating clinician); OS was defined as treatment start to death for any reason. Univariate and multivariable Cox regression analyses served to determine predictive variables. All analyses were performed by IBM SPSS Statistics and R statistics. Data analysis was performed on June 21, 2020.

Fifty-five eligible patients were identified. All patients were followed for more than 3 months with a median follow-up time of 14 months. The median age of patients was 63 years (range 23–83), 69% had M1c or d disease.

Pre-existing autoimmune diseases included Crohn’s disease (N=3), ulcerative colitis (N=7), Graves’ disease (N=8), Hashimoto’s thyroiditis (N=11), RA (N=7), psoriasis (N=6), multiple sclerosis (N=2) and others (table 1). Three patients (5%) had more than one autoimmune disease. At commencement of combination immunotherapy, 10 patients (18%) had active symptoms of their autoimmune condition. Thirteen patients (24%) were receiving immunosuppressive therapy (five on corticosteroids, five on steroid-sparing agents and three on both).

Patients received ipilimumab in combination with either nivolumab (N=46) or pembrolizumab (n=9). Forty patients (73%) received an ipilimumab dose of 3 mg/kg and nivolumab 1 mg/kg 3 weekly for four doses followed by maintenance nivolumab 3 mg/kg. Fifteen (27%) had an alternate regimen with a lower dose of ipilimumab of 1 mg/kg; six with nivolumab 3 mg/kg every 3 weeks for four doses followed by maintenance nivolumab, and nine with pembrolizumab 2 mg/kg every 3 weeks for four doses followed by maintenance pembrolizumab. Anti-PD1 monotherapy was given previously to nine patients (16%), resulting in no flares of autoimmune disease and three (33.3%) patients had experienced previous grade 1 or 2 irAEs.

Eighteen patients (33%) experienced a flare of their autoimmune disease (figure 1, table 2), most often in rheumatic and gastrointestinal disorders. This included 4 of 7 with RA, 3 of 6 with psoriasis, 3 of 7 patients with ulcerative colitis, (including one of two who had undergone previous subtotal colectomy), 2 of 3 with Crohn’s disease, 1 of 8 with Graves’ disease, 2 of 11 with Hashimoto’s thyroiditis, 1 of 1 with Sjogren’s syndrome, 1 of 1 with polymyalgia rheumatica, and 1 of 1 with concurrent Bechet’s syndrome and psoriasis. Flare of autoimmune conditions was based on clinical history and clinician diagnosis. For the patients with RA and Sjogren’s syndrome, this was confirmed with serological testing. For patients with a flare of IBD, four of the five patients who experienced a flare underwent a biopsy for histopathological confirmation. The median time to flare was 19 days (range 4–167). The median time to flare for rheumatologic conditions was 16 days and for IBD was 28 days. Sixteen of the 18 (89%) of the flares occurred during combination therapy, and 2 of 18 (11%) occurred on single-agent PD1 maintenance treatment.

Thirteen of the 18 (72%) patients with a flare were managed with corticosteroids and six (33%) required additional immunosuppressive agents which included ciclosporin, sulfasalazine, infliximab, methotrexate and leflunomide. Seven patients (39%) were hospitalized for management of a flare (three with ulcerative colitis, two with Crohn’s disease and two with RA). Two patients (11%) required admission to intensive care and vasopressor support for severe flares of ulcerative colitis, which had been quiescent prior to treatment and off immunosuppression. One patient suffered diarrhea and shock and the other a duodenal perforation, both recovered and were responsive to intravenous steroids and sulfonamides. Five patients (28%) permanently ceased both drugs due to the flare (two with IBD, one with RA, one with Behcet’s syndrome, one with Graves disease). The median number of doses administered prior to cessation was 1 (range 1–8). Eight of the 18 (44%) patients ceased either both immunotherapy agents or continued on anti-PD1 agent alone due to the flare.

There was a numerical trend to flare of autoimmune disease in patients who had clinically active disease more than those who had clinically inactive disease (50% vs 29%, OR 1.59; 95% CI 0.34 to 7.38, p=0.56). There were also more flares in patients on immunosuppression than in those not on immunosuppression for their autoimmune disorders at immunotherapy commencement (39% vs 26%, OR 4.59; 95% CI 1.16 to 18.04, p=0.03). Notably, in the patients with RA with severe flare (defined as severe pain, joint swelling and limiting self-care), all had quiescent symptoms prior to commencement of therapy. In the four patients with IBD who had a severe flare, none were on immunosuppression and all had been in clinical remission for greater than 24 months prior to commencement of combination immunotherapy.

Thirty-seven patients (67%) experienced an irAE unrelated to their autoimmune disease (table 3). Twenty-one patients (38%) had grade 3–4 irAEs, these included colitis (N=9), hepatitis (N=5), colitis and hepatitis (N=2), pneumonitis (N=2), thyroiditis (N=1), myasthenia gravis (N=1) and GBS (N=1). Nine patients (17%) experienced both flare of autoimmune disease and an irAE. All patients with a grade 3 or 4 immune-related colitis underwent diagnostic colonoscopy or flexible sigmoidoscopy and histopathological confirmation.

Twenty-five patients (68%) required immunosuppression for the management of irAE. Seven patients were managed with oral steroids, seven with intravenous steroids, and nine patients were managed with steroids and steroid-sparing agents; one patient with GBS was managed with intravenous immunoglobulin and one patient with myasthenia gravis was treated with plasmapheresis. Both drugs were ceased due to irAE in 15 patients (41%).

A further subanalysis was undertaken to compare the outcomes of patients who had a flare of their autoimmune disorder with patients who had an irAE alone or both an irAE and a flare. Ten patients had no flare or irAE, 27 patients had an irAE alone, 9 patients had a flare of their autoimmune disorder alone, and 9 had both a flare and an irAE (online supplemental table 1). There was a slightly higher use of immunosuppression to manage patients with both a flare and irAE (89%) compared with irAE alone (70%) or flare alone (78%). More patients who experienced both a flare and irAE were alive at the time of analysis (78%) compared with no irAE or flare (60%), irAE alone (70%) and flare alone (67%). Unfortunately the small numbers in this subanalysis limit statistical evaluation.

A partial or complete response was observed in 30 patients (55%), 30% partial and 25% complete. Seventy-seven per cent of responses are ongoing, with a median follow-up time of 14 months (figures 1 and 2). The objective response rate was not statistically different in those who had a flare of their autoimmune disease versus those who did not (44% vs 59%, p=0.39) nor in patients who were on baseline immunosuppression versus not (46% vs 57%, p=0.72).

Fourteen patients had a complete response (online supplemental table 2). All of these patients had a performance status of ECOG 0. At commencement of combination immunotherapy, two of the complete responders (14%) had active symptoms of their autoimmune condition and were receiving corticosteroids. Three (21%) of these patients experienced a flare of their autoimmune diseases (one with ulcerative colitis, one with psoriasis, one with polymyalgia). Twelve (86%) patients experienced an irAE with seven of these being grade 3 or 4 (online supplemental table 3). Eleven patients remain in complete response.

The median PFS was 10 months (95% CI 0.01 to 32.10) (figure 3A). Patients with flare of their autoimmune disease had a median PFS of 5 months compared with 11 months in those without a flare (p=0.73) (figure 4A). At 12 months, PFS rates were 33% vs 22% in those with a flare of their autoimmune disease versus no flare, respectively. Patients on baseline immunosuppression had a median PFS of 3 months (95% CI 0.71 to 5.30) compared with 33 months (95% CI 8.38 to 57.62) for patients not on baseline immunosuppression (p=0.012) (figure 5A). At 12 months, PFS rates were 15% vs 48% in those on immunosuppression versus no immunosuppression, respectively.

At the time of analysis, a total of 17 patients (31%) had died. One patient was lost to follow up following progression of disease and was not included in the OS analysis. Median OS was 24 months (95% CI 15.88 to 32.12) (figure 3B). Median OS in patients who had a flare of their autoimmune disease was 19 months vs 24 months in those who did not (p=0.37) (figure 4B). Twelve-month OS rates were 39% in patients who had a flare of autoimmune disease vs 64% in those who did not. Median OS in patients who were on baseline immunosuppression was 11 months (95% CI 3.42 to 18.58) compared with 31 months (95% CI 20.89 to 41.11, p=0.005) (figure 5B). Twelve-month OS rates were 23% in patients on immunosuppression vs 66% in those who were not.

Given there were a significant number of patients with an underlying endocrine autoimmune condition, a further subanalysis was performed to assess the difference in outcomes (online supplemental table 4). Endocrine autoimmune disorders were defined as endocrine only disorders and not patients who had an additional autoimmune condition (18 with thyroiditis, 1 with type 1 diabetes mellitus). Patients with an endocrine autoimmune disorder in patients versus patients with non-endocrine autoimmune disorders were less likely to have a flare 16% vs 42% (two-sided Fisher’s exact p=0.05). The rates of irAEs were similar across both groups. The median OS was worse in the patients with a non-endocrine autoimmune disorder at 22 months vs 35 months in patients with an endocrine autoimmune disorder (p=0.046).

Nineteen patients (35%) were given subsequent therapy following combination immunotherapy, some of these had multiple lines of therapy. Alternate anti-PD1 therapy was given in 11 patients (58%), BRAF-MEK inhibitors in 8 patients (42%), 1 patient (5%) was treated with chemotherapy and 5 patients (26%) were enrolled on clinical trials. Of the patients given subsequent therapies, 58% had a partial or complete response.