Preexisting autoimmune disease and immune-related adverse events associated with anti-PD-1 cancer immunotherapy: a national case series from the Canadian Research Group of Rheumatology in Immuno-Oncology

Abstract

Background

Limited data are available on the safety and efficacy of immune checkpoint inhibitors (ICI) in patients with preexisting autoimmune diseases (PAD).

Methods

Retrospective study of patients with PAD referred for rheumatologic evaluation prior to starting or during immunotherapy between January 2013 and July 2019 from 10 academic sites across Canada. Data were extracted by chart review using a standardized form.

Results

Twenty-seven patients with PAD on ICI therapy were identified. The most common PADs were rheumatoid arthritis (30%), psoriasis/psoriatic arthritis (30%), inflammatory bowel disease (IBD, 15%) and axial spondyloarthritis (11%), and the most frequently observed cancers were lung cancer and melanoma. All patients received anti-PD-1 therapies, and 2 received additional sequential anti-CTLA-4 therapy. PAD exacerbations occurred in 52% over a median (IQR) follow-up of 11.0 (6.0–17.5) months, with 14% being severe, 57% requiring corticosteroids, 50% requiring immunosuppression and 14% requiring ICI discontinuation. Flares were generally more frequent and severe in patients who previously required more intensive immunosuppression (i.e., biologics). Flares occurred despite background immunosuppression at the time of ICI initiation. In patients with preexisting psoriasis, IBD and axial spondyloarthritis, rheumatic immune-related adverse events (irAEs), mostly polyarthritis and tenosynovitis, were frequently observed. Tumor progression was not associated with exposure to immunosuppressive drugs before or after ICI initiation and was numerically less frequent in patients with irAEs.

Conclusion

PAD exacerbations in the context of ICI treatment are common, although generally mild, and occur despite background immunosuppression. Exacerbations are more frequent and severe in patients on more intensive immunosuppressive therapies pre-immunotherapy.

Introduction

Immune checkpoint inhibitors (ICI) have revolutionized the treatment of many advanced-stage malignancies. Current ICI target cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), programmed cell death protein 1 (PD-1) and its ligand (PD-L1), releasing the brakes on immune regulatory pathways, termed immune checkpoints, resulting in enhanced immune stimulation and anti-tumor immunity [1]. However, as a result of this immune activation, ICIs are associated with a broad spectrum of off-target immune adverse events, referred to as immune-related adverse events (irAEs). These irAEs can affect most organ systems, with the most common being skin, liver, thyroid and gastrointestinal tract. Although most often mild and manageable, irAEs can be severe and associated with significant morbidity and mortality [1].

Concerns for toxicity are further heightened when considering ICI therapy in patients with preexisting autoimmune diseases (PAD). By unleashing the immune system, ICIs may lead to PAD exacerbations, especially if the drug’s mechanism of action is closely linked to PAD pathophysiology [2]. Patients with PAD may also be predisposed to develop more severe irAEs, given their underlying abnormal immune response against self-antigens. Based on these concerns, patients with PAD were excluded from large randomized controlled trials evaluating the efficacy and safety of ICIs. Since then, there have been limited data from mostly case series/case reports on the course of specific PADs in the context of ICI therapy [2–8].

The Canadian Research Group of Rheumatology in Immuno-Oncology (CanRIO) is a national network of Canadian rheumatologists with interest and expertise in the assessment of patients with rheumatic disease in the context of immunotherapy. In this retrospective study, we report the clinical presentation, management and disease course of patients with PAD treated with ICI at CanRIO sites.

Methods

CanRIO was established in May 2018 by a group of rheumatologists from ten academic centers in Canada. In preparation for the collection of prospective data, the sites pooled data and developed a retrospective cohort of patients referred for rheumatologic evaluation prior to starting or during immunotherapy. All patients were adults and received either a CTLA-4, PD-1 or PD-L1 inhibitor alone or in combination, either in the course of routine clinical care or as part of a clinical trial. Chart review was performed using a standardized data collection form that included demographic information (age, gender, ethnicity), malignancy type and stage, type and duration of immunotherapy, rheumatic and other irAEs classified based on common terminology criteria for adverse events (CTCAE) v5.0 grade [9], irAE treatment and response, and tumor response to ICI based on imaging and impression of the treating oncologist on chart review. Mild flares/irAEs were defined as CTCAE grades 1 or 2, and major/severe flares/irAEs as grades 3 or 4. If PAD was present, disease activity and treatment of PAD before, at and after ICI initiation were recorded. Serologies and imaging were retrieved if available. In this study, immunosuppressants included methotrexate (MTX), hydroxychloroquine (HCQ), sulfasalazine (SSZ), 5-aminosalicylic acid (5-ASA), apremilast, mycophenolic acid, cyclosporine and secukinumab. Descriptive statistics were used to summarize the data. Chi-square tests were used to compare categorical data between groups. Ethics approval for the study was obtained centrally at the Jewish General Hospital, Montreal, Canada, and at individual sites.

Results

Patient characteristics

The CanRIO network identified 27 patients with PAD referred for rheumatologic evaluation prior to starting or during immunotherapy between January 2013 and July 2019. Of these, 19 patients had a rheumatic PAD, and 8 patients had a non-rheumatic PAD and developed rheumatic irAEs. An additional 117 patients without PAD were identified and have been described in detail elsewhere [10]. Of the 27 patients with PAD, 96% were Caucasian and 52% were male with a mean age of 63.6 years (Table 1). Lung cancer and melanoma were the most frequently observed malignancies. Most (93%) patients had stage 4 disease. All patients received anti-PD-1 therapies, and two patients received additional sequential anti-CTLA-4 therapy, for a median (interquartile range, IQR) duration of 4.0 (2.0–9.0) months.

Table 1.

Summary of demographic, autoimmune disease, cancer, treatment and outcomes data

Characteristics [n (%)]	Patients with PAD (n = 27)
Demographics
Age, years (mean, SD)	63.6 (11.5)
Male	14 (52)
Caucasian	26 (96)
Pre-existing autoimmune disease
Rheumatoid arthritis	8 (30)
Psoriasis/psoriatic arthritis	8 (30)*
Inflammatory bowel disease	4(15)*
Axial spondyloarthritis	3 (11)
Systemic lupus erythematosus	2 (7)
Polymyalgia rheumatica	1 (4)
Dermatomyositis	1 (4)*
Others	3 (11)
Cancer type
Lung	15 (56)
Melanoma	8 (30)
Other	4 (15)
Cancer stage
2	1 (4)
3	1 (4)
4	25 (93)
Cancer therapy
Pembrolizumab	20 (74)
Nivolumab	7 (26)
Ipilimumab	2 (7)*
Duration of ICI therapy, months (median, IQR)	4.0 (2.0–9.0)
Immune-related adverse events
No irAE	6 (22)
Any irAE	21 (78)
Flares	14 (52)
New irAE	14 (52)
New rheumatic irAE	10 (37)
New other irAE	6 (22)
CTCAE grade (PAD flares; new irAEs)
1 (asymptomatic/mild)	4 (15); 3 (11)
2 (moderate)	8 (30); 6 (22)
3 (severe)	2 (7); 4 (15)
4 (life-threatening)	0; 1 (4)
Time to flare, months (median, IQR)	< 1 (< 1–2)
Treatment for irAE (for PAD flare; for new irAE)
Prednisone > 10 mg daily	8 (30); 10 (37)
Immunosuppressive drugs	7 (26); 9 (33)
Biologic drug	1 (4); 0
ICI discontinuation (due to PAD flare; due to new irAE)	2 (7); 8 (30)
Tumor response to ICI therapy*
Complete response	4 (15)
Partial responses	10 (37)
Stabilization	1 (4)
Progression	11 (41)
Follow-up time, months (median, IQR)	11.0 (6.0–17.5)

ICI immune checkpoint inhibitor; irAE immune-related adverse event; PAD preexisting autoimmune disease

*Three patients had two pre-existing autoimmune diseases. All psoriatic arthritis patients had psoriasis. One patient received ipilimumab followed by pembrolizumab, and one patient received pembrolizumab followed by ipilimumab. Eleven patients had more than 1 irAE. Data on tumor response were missing in 1 patient

The most common PADs were rheumatoid arthritis (RA, 30%), psoriasis and/or psoriatic arthritis (30%), inflammatory bowel disease (IBD, 15%), axial spondyloarthritis (11%) and systemic lupus erythematosus (SLE, 7%). In addition, polymyalgia rheumatica (PMR), dermatomyositis, interstitial pneumonia with autoimmune features, cutaneous lupus erythematosus and Graves’ disease were observed in one patient each. Among 16 patients on systemic therapies prior to ICI initiation for the treatment of their PAD, 7 (44%) had their therapies continued (including 3 on HCQ, 2 on 5-ASA, one on SSZ/apremilast and one on MTX/HCQ), 7 (44%) had their treatments stopped or reduced after cancer diagnosis (mostly biologics and MTX), and 2 (13%) had their treatments intensified or resumed for flare prevention or treatment (by adding MTX). In addition, 3 patients with axial spondyloarthritis were treated with non-steroidal anti-inflammatory drugs (NSAID) alone, and 6 patients with non-rheumatic PAD as well as two patients with seronegative non-erosive arthritis who were in remission or low disease activity required no systemic therapies.

Flares of preexisting autoimmune diseases

Flares of PAD occurred in 14 (52%) patients over a median (IQR) follow-up of 11.0 (6.0–17.5) months, with 2 (14%) being severe, 8 (57%) requiring corticosteroids, 7 (50%) requiring immunosuppressive drugs and 2 (14%) requiring ICI discontinuation (Table 1). Flares occurred quickly, with the majority (57%) occurring within the first month of ICI initiation.

Rheumatoid arthritis

Among 8 patients with RA, 4 (50%) experienced grade 2 arthritis flares (Table 2). Two of these patients (with RF+, erosive and longstanding RA) were previously treated with biologics (adalimumab and tocilizumab, stopped at the time of cancer diagnosis, 3 and 4 years prior to ICI initiation) and flared despite continuation of MTX in both patients, and preventive addition of HCQ and SSZ in one patient. A third patient with RF+/CCP+ non-erosive RA who was in remission for the past 5 years had MTX stopped at the time of cancer diagnosis, 8 months prior to ICI initiation. MTX was resumed 1 month prior to ICI treatment, but did not prevent a flare which occurred within one dose of ICI. On the other hand, 4 other patients, including two with RF+/CCP+ RA who had MTX stopped 4–9 months prior to ICI initiation, did not experience any arthritis flare.

Table 2.

Flare rates and new irAEs according to PAD type, treatment and disease activity

Characteristics	PAD flares		New irAEs
	Grade 1–2	Grade 3–4	Grade 1–2	Grade 3–4
Rheumatic PAD
Rheumatoid arthritis (n = 8)	4 (50%)	0	2 (25%)	0
No therapy; remission, RF-/CCP-, non-erosive (n = 1)	1; 8 m		0
MTX/HCQ continued; remission, erosive (n = 1)	0		0
HCQ continued, MTX added for flare; RF-/CCP-, non-erosive (n = 1)	0		0
MTX/HCQ stopped for cancer treatment (-4 m); LDA, RF + /CCP + , diagnosed since 3y (n = 1)	0		1; PMR/colitis, < 1 m
MTX/HCQ stopped at cancer diagnosis (-9 m); remission, RF + /CCP + , non-erosive (n = 1)	0		0
MTX stopped at cancer diagnosis (-8 m), resumed 1 m pre-ICI; remission × 5 years, RF + /CCP + , non-erosive (n = 1)	1; < 1 m		1; colitis, 14 m
Tocilizumab stopped at cancer diagnosis (-4y), MTX continued, HCQ/SSZ added 1 m pre-ICI; LDA, RF + , erosive, longstanding disease (33y) (n = 1)	1; < 1 m		0
Adalimumab stopped at cancer diagnosis (-3y), MTX continued; LDA, RF + , erosive, longstanding disease (23y) (n = 1)	1; < 1 m		0
Psoriatic arthritis / peripheral SpA (n = 4)	3 (75%)	0	1 (25%)	1 (25%)
No therapy; LDA (n = 1)	1; < 1 m		1; hepatitis, colitis, 4 m	0
MTX stopped at cancer diagnosis (-1 m); LDA (n = 1)	0		0	0
MTX stopped at cancer diagnosis (-3y), HCQ/SSZ continued; remission (n = 1)	No flare on Pembro; flare at < 1 m on Ipi		0	0
Apremilast stopped within 1 week of ICI initiation, SSZ/Prednisone 5 mg continued; high disease activity (n = 1)	1; < 1 m (after stopping Apremilast)		0	1; hepatitis, < 1 m (before stopping Apremilast)
Axial spondyloarthritis (n = 3): NSAID continued; inactive/LDA	0	0	1 (33%); polyarthritis,  < 1 m	1 (33%); polyarthritis, 8 m
SLE (n = 2): HCQ continued; remission	0	0	0	0
PMR (n = 1): HCQ continued; remission	1; < 1 m	0	0	0
Dermatomyositis (n = 1): no therapy; remission	0	0	0	1; neutropenia, 1 m
Non-rheumatic PAD
Psoriasis (n = 7)	4 (57%)	2 (29%)	3 (43%)	1 (14%)
No therapy; remission/LDA (n = 3)	3; < 1 m	0	1 polyarthritis, 1 m; 1 hepatitis, colitis, 4 m	1; polyarthritis, tenosynovitis, < 1 m
Topical therapy; stable severe psoriasis (n = 1)	0	1; 4 m	1; arthritis, sacroiliitis, 4 m	0
MTX stopped at cancer diagnosis (-3y); remission (n = 1)	0	0	0	0
MTX stopped at cancer diagnosis (-1 m); flare (n = 1)	1; 3 m	0	0	0
Ustekinumab/MTX stopped after cancer diagnosis (-6 m), apremilast continued; stable erythrodermic psoriasis (n = 1)	0	1; 2 m	0	0
Inflammatory bowel disease (n = 4)	0	1 (25%)	1 (25%)	2 (50%)
No therapy, remission (n = 1) (+ celiac)	0	0	1; polyarthritis, mucositis, < 1 m	0
5-aminosalicylic acid continued, remission (n = 2) (+ psoriasis) (+ dermatomyositis)	0	0	0	1 polyarthritis, tenosynovitis, < 1 m; 1 neutropenia, 1 m
Ustekinumab/MTX stopped after cancer dx (-6 m) (n = 1) (+ psoriasis)	0	1; 2 m	0	0
Interstitial pneumonia with autoimmune features (n = 1); no therapy, new diagnosis	1; 2 m	0	1; sicca, 2 m	0
Cutaneous lupus (n = 1); no therapy, remission	0	0	0	1; polyarthritis-tenosynovitis, 1 m
Graves' disease (n = 1); no therapy, remission	0	0	1; acral vasculitis, myositis, < 1 m	0

CCP cyclic citrullinated peptide; HCQ hydroxychloroquine; Ipi ipilimumab; irAE immune-related adverse event; LDA low disease activity; m month(s) from ICI initiation; MTX methotrexate; NSAID non-steroidal anti-inflammatory drug; PAD preexisting autoimmune disease; Pembro pembrolizumab; PMR polymyalgia rheumatica; RF rheumatoid factor; SLE systemic lupus erythematosus; SSZ sulfasalazine; y year(s) from ICI initiation

RA flare responded to increased oral MTX dosage alone (from 15 to 20 mg per week) in one patient. Two patients responded to moderate-dose prednisone (25–30 mg daily), but one patient remained steroid-dependent despite continuous use of MTX/HCQ/SSZ (past tocilizumab user), while another patient achieved remission with additional use of MTX/HCQ/SSZ and ICI discontinuation (previously on no therapy). The fourth patient (past adalimumab user) was treated with intra-articular corticosteroids and addition of SSZ to MTX, but this did not prevent worsening of the flare on subsequent ICI infusions.

Psoriatic arthritis

Among 4 patients with psoriatic arthritis, 3 (75%) experienced grade 2 arthritis flares. In one patient, the arthritis flared after apremilast was discontinued in the context of a grade 3 hepatitis that occurred after the first dose of ICI therapy, despite continued use of SSZ and prednisone 5 mg daily. In another patient, the arthritis remained quiescent during 3 months of pembrolizumab therapy, but flared after one dose of ipilimumab (changed for tumor progression) despite background HCQ/SSZ therapy. Psoriatic arthritis flares responded to NSAIDs alone in one patient and to addition of MTX along with high-dose prednisone (50 mg daily) and ICI discontinuation (for concomitant colitis) in another patient. HCQ alone was insufficient to treat arthritis flares in two patients.

Other rheumatic preexisting autoimmune diseases

One patient with preexisting PMR, diagnosed 10 years prior to ICI initiation and in remission on HCQ therapy (which was continued during ICI therapy), had a mild flare after one dose of ICI, which responded to dexamethasone 1 g daily (high dosage given for concomitant cancer-related dyspnea). None of the 6 patients with axial spondyloarthritis, SLE or dermatomyositis had flares of their underlying rheumatic disease.

Non-rheumatic preexisting autoimmune diseases

The majority (86%) of patients with psoriasis had flares of their skin disease, including grade 3 flares in 2 patients with previously severe disease. One patient, previously treated with apremilast, MTX and ustekinumab for erythrodermic psoriasis and Crohn’s disease, had the two latter immunosuppressants discontinued at the time of cancer diagnosis over 6 months prior to ICI initiation and developed a grade 3 psoriasis flare and a grade 2 colitis flare after 3 doses of pembrolizumab. This patient was treated with secukinumab with complete resolution of skin and gastrointestinal symptoms, but had a loss of tumor response to the ICI, as evidenced by an increase in carcinoembryonic antigen (CEA) levels on secukinumab after an initial decrease on ICI. The patient died 11 months after ICI initiation due to tumor progression [11]. In another patient with severe but stable psoriasis on topical therapy, a major psoriatic flare occurred with 80–90% body surface area involvement, which markedly improved on high-dose prednisone (50 mg daily) and MTX. The ICI was held for 2 weeks and resumed without any further adverse events. Three other patients with psoriasis who were in remission or low disease activity on no therapy experienced mild flares of their skin disease; two were effectively treated with MTX. Another patient with severe psoriasis had active disease at the time of ICI initiation (following discontinuation of MTX). The psoriasis continued to worsen with ICI therapy, which was discontinued after 4 months due to tumor progression. Only one patient did not experience any flare in skin disease; this patient had MTX stopped 3 years prior to ICI treatment, at the time of cancer diagnosis.

New immune-related adverse events

New irAEs occurred in 14 (52%) patients, of which 5 (36%) were grade 3–4 events, 10 (71%) required corticosteroids, 9 (64%) required immunosuppression, and 8 (57%) required ICI discontinuation (Table 1). Rheumatic irAEs were frequently observed among patients with preexisting psoriasis, IBD and axial spondyloarthritis (Table 2). In fact, out of 4 psoriasis, 4 IBD and 3 axial spondyloarthritis patients without arthritis at baseline, 3 (75%), 2 (50%) and 2 (66%) developed de novo arthritis within 1–8 months of ICI initiation, respectively. New arthritis was severe in 2 (33%) patients and characterized as symmetric polyarthritis (67%) with flexor tenosynovitis (33%), asymmetric polyarthritis (17%) or psoriatic-like arthritis (17%). Arthritis responded to prednisone 30–50 mg daily in 4 patients (higher dosages were used due to other concomitant irAEs). MTX was effective as a steroid-sparing agent in 4 patients, whereas HCQ and SSZ were not effective in 2 patients.

One patient with dermatomyositis (diagnosed 11 years earlier and in remission) and Crohn’s disease (in remission on 5-aminosalicylic acid) had quiescent PAD but developed grade 4 neutropenia after two doses of pembrolizumab, which recovered after 2 weeks of intravenous and high-dose oral corticosteroids, granulocyte colony-stimulating factor, cyclosporine, intravenous immunoglobulins and ICI discontinuation thereafter [12].

One patient with Graves’ disease developed acral vasculitis and asymptomatic myositis (creatine kinase greater than 2 times the upper limit of normal and muscle inflammation on MRI) within 1 week of starting pembrolizumab. Serology revealed a positive ANA (1:640 homogenous) and anti-Jo-1 antibody, and this patient required very high doses of corticosteroids to control the ischemic lesions. In fact, after 3 days of intravenous methylprednisone (IVMP) at 1000 mg daily, the lesions worsened on prednisone 50 mg and IVMP 125 mg for 3 days, leading to re-treatment with IVMP (500 mg for 3 days) and then prednisone 100 mg for 3 days. Lesions subsequently improved, prednisone was slowly tapered, and mycophenolic acid was added.

One patient with non-specific interstitial pneumonitis with positive ANA and Ro52 autoantibodies developed mild sicca and asymptomatic progression in ground-glass attenuations at a follow-up CT scan 2 months after ICI initiation. He was treated with a 3-week course of prednisone (50 mg daily). The ICI was held during that treatment period then resumed. The interstitial pneumonitis continued to progress at 6 months post-ICI initiation but then stabilized at 11 months while remaining on the ICI.

Cancer outcomes

Of the 27 patients with PAD, 4 (15%) achieved complete response, 10 (37%) had a partial response, 1 (4%) had tumor stabilization, and 11 (41%) experienced tumor progression with ICI therapy (Table 1). Mean (SD) follow-up was 13.8 (7.6) and 12.7 (11.2) months among patients with and without tumor progression.

Cancer outcomes according to irAE and treatment characteristics are outlined in Table 3. The proportion of patients exposed to immunosuppressive drugs at the time of ICI initiation was not significantly different between patients with and without tumor progression (55% vs. 47%, p = 0.69). Patients with tumor progression numerically less frequently developed irAEs (73% vs. 87%, p = 0.37 for any irAE; 27% versus 53%, p = 0.18 for at least two irAE), but numerically more frequently had irAEs leading to ICI discontinuation (45% vs. 33%, p = 0.53). After ICI initiation, patients with tumor progression were numerically less frequently exposed to corticosteroids (36% vs. 73%, p = 0.06) and were similarly exposed to immunosuppressive drugs (82% vs. 87%, p = 0.73) for the treatment of irAEs, compared to those without tumor progression. However, in one patient, the positive response achieved on ICI was lost after secukinumab was initiated for management of a major psoriasis and colitis flare. Out of 10 patients in whom ICI was discontinued due to an irAE, 5 experienced tumor progression, including 3 who had an initial partial response or stabilization on ICI, with loss of tumor response after ICI discontinuation. Conversely, partial or complete responses were achieved in 5 patients despite ICI discontinuation.

Table 3.

Cancer outcomes according to immune-related adverse events and treatment characteristics*

Characteristics [n (%)]	No progression (n = 15)	Progression (n = 11)
PAD type—rheumatic
Rheumatoid arthritis	2/8	6/8
Psoriasis	4/7	3/7
Psoriatic arthritis/peripheral SpA	2/4	2/4
Inflammatory bowel disease	2/4	2/4
Axial spondyloarthropathy	2/2	0/2
SLE	2/2	0/2
Systemic therapies for PAD at ICI initiation
Any Prednisone	1 (5 mg)	0
Any immunosuppressive drug	7 (47%)	6 (55%)
MTX	1	1
MTX/HCQ	0	2
MTX/HCQ/SSZ	1	0
HCQ	3	0
HCQ/SSZ	0	1
5-aminosalicylic acid	1	1
Apremilast	0	1
Apremilast/SSZ/Pred 5 mg	1	0
Immune-related adverse events
Any irAE	13 (87%)	8 (73%)
Number of irAE
1	5	5
2	5	3
3	2	0
4	1	0
Maximal irAE grade
1	2	0
2	7	5
3	4	2
4	0	1
Systemic therapies after ICI initiation
Any corticosteroids*	11 (73%)	4 (36%)
Low-dose oral corticosteroids (5 mg)	1 (new arthritis at < 1 m)
Moderate-dose oral corticosteroids (25–30 mg)	2 (arthritis flare at < 1 m; new arthritis at 8 m)	1 (arthritis flare at 8 m)
High-dose oral or intravenous corticosteroids (≥ 50 mg)	8 (psoriasis flare and new arthritis [< 1 m, 4 m], PMR flare [< 1 m], new hepatitis [< 1 m], new acral vasculitis [< 1 m], new colitis [5 m], ILD progression [2 m], mucositis/gastritis [3y])	3 (new PMR at < 1 m, new arthritis at 1 m, new neutropenia at 1 m)
Any immunosuppressive drug	13 (87%)	9 (82%)
MTX	2 (2 added for psoriasis flare and new arthritis)	2 (1 continued, 1 added for new PMR)
MTX/HCQ	1 (added for arthritis flare)	2 (2 continued)
MTX/HCQ/SSZ	3 (1 continued, 2 added for new arthritis)	1 (added for arthritis flare)
MTX/SSZ	1 (MTX continued, SSZ added for arthritis flare)	0
HCQ/SSZ	1 (SSZ continued, HCQ added for arthritis flare)	1 (continued)
HCQ	4 (3 continued, 1 added for new arthritis)	1 (added for new arthritis)
Mycophenolic acid	1 (added for new acral vasculitis and myositis)	0
Cyclosporin/IVIg	0	1 (added for new neutropenia)
Secukinumab	0	1 (added for psoriasis and colitis flare)
ICI discontinuation for irAE	5 (33%)	5 (45%)
For flare	0	2
For new irAE	5	3

HCQ hydroxychloroquine; ICI immune checkpoint inhibitor; irAE immune-related adverse events; m month(s) from ICI initiation; MTX methotrexate; PAD preexisting autoimmune disease; SSZ sulfasalazine

*Data on tumor response were missing in 1 patient. Corticosteroid doses are in prednisone-equivalent daily dose

Discussion

In this Canadian national case series of 27 patients with PAD referred for rheumatologic evaluation in the context of mainly anti-PD-1 ICI therapy, the most commonly observed PAD were RA, psoriasis/psoriatic arthritis, IBD and axial spondyloarthritis. The majority (78%) of patients developed at least one irAE, which were generally mild events, but were severe leading to ICI discontinuation in a third of irAEs. PAD exacerbations were common in patients with RA (50%) and psoriatic arthritis (75%) despite no or low disease activity and background immunosuppression at the time of ICI initiation, often occurred early after ICI initiation and were more difficult to manage in the two patients who previously required biologic treatments prior to cancer diagnosis. In patients with preexisting psoriasis, flares of skin disease occurred in the majority of patients (86%) and were most severe among patients with preexisting severe psoriasis requiring systemic therapies. Peripheral arthritis was frequently observed as a new irAE among patients with psoriasis, IBD and axial spondyloarthritis.

These findings are consistent with results reported from other retrospective case series evaluating PAD subjects exposed to PD-1/PD-L1 ICI. In the series by Tison et al. (n = 112), flares were also observed in 47% of patients, with high flare rates reported in RA (60%) and psoriasis (68%) [3]. In contrast, other case series have reported flare rates of only 23% to 38% [5–8], although flares were more frequent among patients with rheumatic PAD (40–54%) [5–7], thus potentially explaining the difference in flare rates between those studies and ours. It is unclear why flares would be more common in rheumatic PAD; however, PD1/PD-L1 pathways play a role in the pathophysiology of several rheumatic autoimmune diseases, including RA, SLE, Sjögren’s syndrome, myositis and ankylosing spondylitis [13–16]. The relatively higher flare rates reported in our series may also be due to referral bias, as patients with more severe rheumatic disease may have been more likely referred for rheumatologic assessment in the context of immunotherapy.

Furthermore, over 40% of patients with PAD experienced tumor progression over a median follow-up of 11 months. In comparison, among patients from the CanRIO cohort without PAD, only 15% of patients progressed over a mean follow-up of 17 months [10]. Patients without tumor progression numerically had more irAEs, which is consistent with previous studies, suggesting that irAEs are predictive of an effective anti-tumor response [2]. Corticosteroids and non-biologic immunosuppressive drugs given for the treatment of irAE after ICI initiation did not appear to attenuate the anti-tumor response, although this association may be confounded by the presence of irAE. The higher progression rate observed in patients with PAD compared to non-PAD patients could not be explained by the rate of severe irAE leading to ICI discontinuation, which was numerically lower in the PAD group (33% vs. 52%) [10]. Alternatively, this higher progression rate may be explained by channeling bias, whereby patients with PAD requiring immunosuppressive drugs and with better prognosis cancers may preferentially be treated with non-ICI cancer therapies in order to avoid PAD flares. If there is a higher threshold of poor cancer prognosis required for clinicians to initiate ICI therapy in these patients, then this may explain why the rate of cancer progression is higher among PAD patients compared to non-PAD patients in this retrospective cohort.

Another possible hypothesis to explain the higher progression rate among patients with PAD is that this group is more often exposed to immunosuppression at the time of ICI initiation, which could blunt immune stimulation by ICI, thus reducing its efficacy against cancer. In the study by Tison et al. shorter median progression-free survival was observed among patients receiving immunosuppression at ICI initiation on a multivariate analysis adjusted for sex, age, type of cancer and type of ICI (HR 2.10, 95% CI: 1.08–4.07) [3]. Similarly, the study by Menzies et al. reported lower response rates among patients on immunosuppression at ICI initiation, adjusted for cancer stage, brain metastases, performance status and/or LDH (15% vs. 44%, p = 0.029) [6]. Steroid use at prednisone-equivalent doses of ≥ 10 mg at or near the time of ICI initiation has also been associated with poor cancer outcomes [17, 18]. On the other hand, in our study, the proportion of patients exposed to immunosuppressive drugs at the time of ICI initiation was not significantly higher among patients with tumor progression.

Interestingly, a patient with preexisting interstitial lung disease (ILD) with anti-Ro52 autoantibodies developed grade 1 sicca as a new irAE 2 months after ICI initiation, in addition to exacerbation of the ILD. Sicca has been described as an irAE secondary to ICI, but with features distinct from Sjögren’s syndrome. In a case series of 20 patients with new or worsening xerostomia after ICI, anti-SSA was only detected in two patients [19]. Salivary gland biopsies in patients with ICI-induced sicca demonstrated mild-to-severe sialadenitis with diffuse lymphocytic infiltration composed mainly of CD3+ T cells, in contrast to the lymphocytic infiltration in Sjögren’s syndrome which is composed mainly of B cells. In patients with profound sialadenitis post-ICI, lymphocytes also exhibited anti-PD-L1 positivity. As salivary gland biopsy was not performed in our patient, it remains unclear whether the ICI directly caused the sicca, unmasked a subclinical preexisting autoimmune condition or triggered Sjögren’s syndrome as the ultimate “hit” in the context of preexisting anti-Ro52 autoantibodies.

We were unable to evaluate whether patients with PAD developed more irAE compared to patients without PAD, given that patients without PAD in the CanRIO cohort were referred for rheumatologic assessment based on rheumatic symptoms arising in the context of immunotherapy, likely leading to an overestimation of the incidence of irAE within this population. In the prospective registry study of grade ≥ 2 irAE by Danlos et al., irAE developed more frequently (44% vs. 24%) and more rapidly (median 5.4 versus 13 months, p = 0.0002) in 45 patients with PAD compared to 352 patients without PAD [8]. Similarly, in a retrospective study by Cortellini et al., the incidence of irAE of any grade was higher among 85 consecutive patients with PAD compared to 666 patients without PAD (66% vs. 40%), although the incidence of grade 3–4 irAE was not different between these groups (9.4% vs. 8.8%) [4]. A study by Abu-Sbeih et al. also reported higher rates of gastrointestinal irAE among 102 patients with preexisting IBD compared to similar patients without this PAD (41% vs. 11%) [20]. Preexisting ILD has also been reported as an independent risk factor for ICI-induced pneumonitis [21], occurring in 30% versus 10% of patients with and without underlying ILD [22, 23].

Our study has limitations inherent to its retrospective design, including selection/referral bias. Patients with more severe rheumatic disease may have been more likely to be referred for rheumatologic assessment. In addition, certain variables were more difficult to ascertain retrospectively, such as disease activity of the PAD prior to ICI initiation if the patient was followed in another center before ICI initiation. Finally, this study only included 27 patients, with heterogeneity in underlying PAD, tumor type, immunotherapy received, dose and type of immunosuppression, making it difficult to generalize to a specific population or therapeutic agent. The small sample size precluded adjusted analyses to explore the association between immunosuppression exposure and risk of tumor progression.

The strengths of this study include detailed review of the clinical evolution of patients by specific PAD, using data gathered from a nationwide multicenter network. Our conclusions are consistent with the National Comprehensive Cancer Network (NCCN) guidelines, which recommend considering ICI in patients with low level of or no immunosuppression with good control of the PAD, but to avoid ICI in patients with neurologic, life-threatening or poorly controlled PAD, or in patients requiring high levels of immunosuppression for PAD control, given that flares in these patients can be difficult to manage and given that immunosuppression at the time of ICI initiation may compromise ICI efficacy [2, 24]. These recommendations were based on a limited number of retrospective case series; our study adds additional evidence to support these guidelines.

Overall, in this retrospective national case series, PAD exacerbations were common, however generally mild and manageable. Exacerbations were more frequent and/or severe in patients who required more intensive systemic therapies pre-ICI and occurred despite preventive addition of immunosuppressive drugs prior to ICI initiation. Nevertheless, the majority of patients with PAD responded to ICI, supporting the current approach to consider ICI treatment in patients with PAD who could benefit from these life-saving therapies. Early rheumatologic referral prior to initiating ICI is essential to ensure that rheumatic irAEs are managed appropriately.

With the increasing use of immunotherapy for more cancers, there are outstanding and evolving questions which need to be addressed with large, prospective, multi-national studies. Should we have the same approach to all patients, regardless of underlying PAD, tumor type, plan for immunotherapy and immunosuppression? Should we try to reduce/minimize immunosuppression in patients with PAD prior to starting ICI? Will this allow for more robust tumor response? Will it increase the risk of irAE? As our experience with cancer immunotherapy increases, we are hopeful that CanRIO and other groups will be able to work together to help clinicians and patients address these complex relationships.

Author contributions

All authors contributed to the study conception and design, or to the acquisition, analysis or interpretation of the data. The first draft of the manuscript was written by Sabrina Hoa and Linda Laaouad, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Compliance with ethical standards

Conflict of interest

The authors have no conflicts of interest to declare that are relevant to the content of this article.

Ethical statement

This submission represents original work that is not under consideration elsewhere and that has not been previously published. All authors participated in the preparation of the manuscript. All authors meet criteria for authorship and will sign a statement attesting authorship, disclosing all potential conflicts of interest and releasing the copyright should the manuscript be acceptable for publication.

Consent to participate

Patient consent was waived due to the retrospective nature of the study.