Cutaneous Immune‐Related Adverse Events and Efficacy of Immune Checkpoint Inhibitors for Patients With Advanced Solid Organ Malignancies

ABSTRACT

Introduction

Immune checkpoint inhibitors (ICIs) have significantly improved outcomes for patients with advanced solid tumors. While low‐grade immune‐related adverse events (irAEs) are associated with prolonged survival, high‐grade irAEs have been associated with poorer survival. Cutaneous immune‐related adverse events (cirAEs) affect up to 20%–40% of patients treated with ICIs. We investigated the association between cirAES and the outcomes of progression‐free survival (PFS) and overall survival (OS) in advanced solid organ malignancies.

Methods

A retrospective analysis of patients receiving ICIs for stage IV solid organ malignancies was conducted at Beaumont RCSI Cancer Centre, Dublin, Ireland, between January 1, 2012, and June 30, 2020. Eligible participants included those who commenced therapy during this period, having received at least one cycle of ICI treatment, with or without chemotherapy, for histologically confirmed advanced solid organ malignancies.

Results

Among 278 analyzed patients, 19% (53/278) experienced any cirAES. The most common cirAEs included psoriasis (23%) and pruritus (15%). cirAES were associated with significantly improved PFS (median 47.3 months vs. 18.3 months, p < 0.01) and OS (median 60.0 months vs. 26.0 months, p < 0.01). Patients with prior systemic therapy had a decreased risk of cirAES (odds ratio = 0.44, p = 0.02), and multivariate analysis confirmed that cirAES was independently associated with improved PFS and OS.

Conclusion

Our study supports that cirAES may be associated with improved patient outcomes and that prior systemic therapy may be associated with a reduced risk of cirAES. Future research should focus on multi‐institutional collaborations based on prospective irAE data to better understand the impact of specific irAEs on clinical outcomes.

1. Introduction

Immune checkpoint inhibitors (ICI) have resulted in significant improvements in outcomes for patients with advanced solid tumors. In the CHECKMATE‐067 study, ipilimumab/nivolumab improved 5‐year survival for patients with advanced melanoma to 52% [1, 2, 3]. Similar outcomes have been seen in malignancies such as non‐small cell lung cancer, renal cell cancer, head and neck cancer, urothelial cancer, and others [4, 5, 6, 7, 8, 9]. Despite these survival benefits across tumor types, a critical challenge is managing immune‐related adverse events (irAEs), which may lead to treatment interruption and cessation.

Whilst most irAEs are mild and self‐limiting, some patients may experience severe, life‐threatening toxicities (Grade 3 or 4). To mitigate these risks, the European Society of Medical Oncology (ESMO) and the National Comprehensive Cancer Network (NCCN) have provided detailed guidelines for managing irAEs [10, 11]. Cutaneous immune‐related adverse events (cirAES) are common and typically encountered in 30% or more of patients [12]. While the majority of cirAEs are mild, they can range from non‐visible toxicities such as pruritus to more severe reactions, including bullous disease and Stevens‐Johnson syndrome [13, 14]. The mechanisms of cirAEs are diverse and dependent on the specific toxicity type, but may include increased levels of IL‐22, IL‐17, depletion of NK cells, and autoantibody development [15]. Management of cirAES secondary to ICIs typically involves a tiered approach based on severity. Mild cases may be treated with antihistamines and low‐potency topical corticosteroids. Moderate to severe cases often require higher‐potency topical corticosteroids or oral corticosteroids. In instances of severe toxicity, high‐dose corticosteroids and other corticosteroid‐sparing strategies may be necessary [10]. The European Academy of Dermatology and Venereology (EADV) has also developed a task force and consensus guidelines, which provide phenotype‐derived management recommendations for patients with cirAEs [16]. Finally, the European Network for Cutaneous Adverse Events to Oncologic Drugs (ENCADO) group has also performed a retrospective review and provided specific guidance on managing ICI‐induced psoriasis [17].

A critical challenge with using ICIs is identifying biomarkers that may predict clinical benefit from treatment to guide treatment decisions, particularly in the context of irAEs. These may include clinical, biochemical, or genomic biomarkers of response and resistance to therapy. It has previously been suggested that there is a positive correlation between the occurrence of irAEs and improved overall survival (OS) in patients with melanoma, non‐small cell lung cancer (NSCLC), and other cancers [18]. High‐grade irAEs, however, such as severe pneumonitis, can lead to worsened OS due to increased morbidity, complications, and potential treatment interruption [19]. cirAES are frequently encountered irAEs but are uncommonly life‐threatening or associated with hospitalization [15]. They therefore may have a potential clinical biomarker of response to ICIs. In a cohort of 128 patients from Taiwan treated with ICIs, the development of irAEs was associated with an improved overall survival (p < 0.01) [20]. Focusing on specific cirAEs, an individual patient‐level systematic review and meta‐analysis identified that the development of vitiligo was associated with an improved PFS [HR = 0.51] and OS [HR = 0.25] [21].

The understanding of clinical risk factors for cirAES remains limited, and further work is needed to elucidate the diversity and clinical impact in a real‐world population. To further define this field, we explored risk factors associated with the development of cirAEs and the association of cirAEs with survival. This included a multivariate model to explore whether survival is independent of other clinical characteristics.

2. Materials and Methods

We performed a retrospective analysis of all patients undergoing ICIs for metastatic (stage IV) solid organ malignancies. We excluded patients with stage III disease, given that most patients have been treated with curative intent. The survival outcomes in this cohort would therefore be a significant confounding variable. Eligible participants commenced therapy between 1st of January 2012 and the 30th of June 2020 at Beaumont RCSI Cancer Centre, Dublin, Ireland. Eligible participants received at least one cycle of treatment with ICI for a histologically proven advanced solid organ malignancy during this time period, regardless of line of therapy. Patients receiving anti‐programmed death ligand 1 (PD‐L1)/anti‐programmed cell death 1 (PD‐1)+/− anti‐ cytotoxic T lymphocyte associated protein 4 (CTLA‐4) were also eligible.

Patients were identified through a review of electronic systemic anti‐cancer therapy records. Data was extracted from medical records. Extracted clinical data included patient demographics, tumor type, prior and current systemic anti‐cancer treatment, cirAEs details (incidence, subtype, and treatment received), PFS, and OS. cirAEs were diagnosed, subtyped, and graded by the treating oncologist and confirmed by a consultant dermatologist when clinically indicated. Data were collected by a medical oncology trainee via chart review. Patients with cirAES were defined as those patients receiving active therapy with ICI or whose most recent treatment was ICI, who developed either a (1) biopsy‐proven cirAES or (2) cirAES that was deemed to be an irAE by the treating medical oncologist +/− dermatologist. The association of cirAEs and survival (OS and PFS) was determined using Cox proportional hazard regression analysis. Analysis was carried out in the R statistical environment using the “survival” and “survminer” libraries (https://cran.r‐projeST.org/).

3. Results

A total of 278 patients who received ICI treatment were included. Among those, 53 (19%) experienced any grade cirAES. The cirAES group had a median age of 62, and 58.5% (31) were male. In the non‐cirAES group, the median age was 62, and 62.2% (140) were female (see Table 1 for details). The most common regimens that patients received were nivolumab (47%, 25/53 in cirAES group; 53%, 123/225 in non‐cirAES group) and pembrolizumab (31%, 16/53 in cirAES group; 30%, 67/225 in non‐cirAES group). In the cirAES group, the most frequent disease types included melanoma (49%, 26/53) and NSCLC (26%, 14/53). In the non‐cirAES group, there was a higher proportion of patients with NSCLC (40%, 90/225) and a lower proportion of patients with melanoma (23%, 51/225). Patient demographic details are outlined in Table 1.

TABLE 1.

Patient demographic details.

	Skin toxicity	No skin toxicity
N = 278	53	225
Males (%)	31 (58.5)	140 (62.2)
Female (%)	22 (41.5)	85 (37.8)
Median age (range)	62 (22, 82)	62 (25, 89)
Prior systemic anticancer treatment	14 (26%)	105 (46%)
Diagnosis
Non‐small cell lung cancer	26% (14/53)	40% (90/225)
Melanoma	49% (26/53)	23% (51/225)
Renal cell carcinoma	11% (6/53)	13% (30/225)
Head and neck cancer	4% (2/53)	14% (31/225)
Urothelial cancer	2% (1/53)	4% (8/225)
Small cell lung cancer	0	2% (5/225)
Other	8% (4/53)	4% (10/225)
ICI Regimen
Nivolumab	47% (25/53)	55% (123/225)
Pembrolizumab	31% (16/53)	30% (67/225)
Ipilimumab + Nivolumab	15% (8/53)	3% (6/225)
Durvalumab	0	4% (8/225)
Atezolizumab	2% (1/53)	5% (11/225)
Pembrolizumab + Chemo	0	3% (7/225)
Ipilimumab monotherapy	8% (4/53)	1% (3/225)
Skin toxicity in overall population
Any skin toxicity	19% (53/278)
Grade 1	15% (8/53)
Grade 2	79% (42/53)
Grade 3	6% (3/53)
Grade 4	0
Skin Toxicity Diagnoses
Pruritus	30% (16/63)
Pruritus with maculopapular rash	15% (8/53)
Pruritus without rash	15% (8/53)
Eczema/Dermatitis	26% (14/53)
Psoriasis	23% (12/53)
Vitiligo	13% (7/53)
Lichenoid Eruption	5% (3/53)
Bullous Pemphigoid	4% (2/53)
Treatment of Skin Toxicity
Anti‐histamines	51% (27/53)
Topical corticosteroids and Antihistamines	33% (17/53)
Oral prednisolone	10% (5/53)
No treatment	8% (4/53)

Abbreviation: chemo, chemotherapy; ICI, immune checkpoint inhibitor.

The most commonly encountered cutaneous irAEs in this analysis were pruritus (30%, 16/53), which was further subdivided into pruritus with a maculopapular rash (15%, 8/53) and pruritus without rash (15%, 8/53), followed by eczema/dermatitis (26%, 14/53), psoriasis (23%, 12/53), vitiligo (13%, 7/53), lichenoid eruption (5%, 3/53), and bullous pemphigoid (4%, 2/53). Most patients were treated with oral antihistamines (51%, 27/53) or topical corticosteroids and antihistamines (33%, 17/53). A small proportion of patients received oral corticosteroids (10%, 5/53) or no treatment (8%, 4/53). No patient received a biologic therapy.

3.1. Univariate Survival Analysis

At the time of data cut‐off, there were 233 PFS (32 in the cirAE group) events and 210 OS (25 in the cirAE group). In the cirAE group, the median duration of follow‐up was 53 months (interquartile range (IQR) 40–70 months). In the non‐cirAE group, the median duration of follow‐up was 58 months (IQR 44–73 months). In the cirAES group, the median PFS was 47.3 months versus 20.9 months in those without cirAEs (hazard ratio (HR) = 0.51, p < 0.01, Figure 1B). Similarly, patients who developed cirAE had a significantly improved OS of 60 months versus 26 months (HR = 0.46, p < 0.01, Figure 1A). We also identified that patients who had one prior line of systemic anti‐cancer therapy had a decreased risk of cirAES (odds ratio = 0.44, 95% confidence interval (CI) 0.22–0.88, p = 0.02).

FIGURE 1.

(A) Overall survival (OS) in those with and without cutaneous toxicity. Patients who developed cutaneous toxicity had a significantly improved OS 60 months versus 26 months (HR = 0.46, p < 0.01, Figure 1A). Events: toxicity group (37%; 20/53); no toxicity group (82%; 185/225). (B) Progression‐free survival (PFS) for those with and without cutaneous toxicity. In the cutaneous toxicity group, the median PFS was 47.3 months versus 20.9 months in those without cutaneous toxicity (HR = 0.51, p < 0.01, Figure 1B). Events: toxicity group (60%; 32/53); No toxicity group (89%; 201/225).

3.2. Multivariate Survival Analysis

In a multivariate analysis (see Table 2), cirAES were associated with an improved PFS [HR = 0.58, p < 0.01] and OS [HR 0.66, p < 0.01] independent of age, sex, disease, and prior systemic anticancer treatment. Of note, even when those with melanoma and renal cell carcinoma were considered as individual subgroups, the development of cirAES was associated with an improved PFS [renal cell carcinoma, HR 0.46, p < 0.01; melanoma HR 0.36, p < 0.01] and OS [renal cell carcinoma, HR 0.44, p < 0.01; melanoma HR 0.38, p < 0.01].

TABLE 2.

Multivariate analysis with adjusted HR for overall survival (OS), progression‐free survival (PFS), and clinical parameters.

Variable	HR [95% CI]	p
PFS
Cutaneous irAEs	0.66 [0.44–0.99]	0.04
Age	1 [0.99–1.01]	0.20
Sex	0.93 [0.69–1.26]	0.67
Melanoma	0.36 [0.25–0.54]	< 0.01
Renal cell carcinoma	0.43 [0.30–0.70]	< 0.01
Head and neck carcinoma	0.74 [0.47–1.16]	0.19
Urothelial cancer	0.71 [0.41–1.23]	0.21
Prior systemic anticancer treatment	1.02 [0.81–1.30]	0.84
OS
Cutaneous irAEs	0.55 [0.36–0.85]	0.006
Age	1 [0.99–1.02]	0.17
Sex	0.85 [0.61–1.17]	0.33
Melanoma	0.38 [0.25–0.58]	< 0.01
Renal cell carcinoma	0.44 [0.26–0.70]	< 0.01
Head and neck carcinoma	0.81 [0.51–1.28]	0.38
Urothelial cancer	0.73 [0.41–1.3]	0.29
Prior systemic anticancer treatment	1.11 [0.86–1.43]	0.40

Note: Patients with cutaneous irAEs (Immune‐related adverse events) had improved OS and PFS independent of age, sex, disease, or prior systemic anticancer therapy.

4. Discussion

In a large cohort of patients treated with ICIs for solid tumors, we identified that those who developed cirAES had a significantly improved PFS and OS compared with those who did not develop cirAES. Notably, patients with cirAES had a median OS of over 4 years and a median PFS of 5 years. This is more than double that of patients who did not develop such toxicities in our cohort. We discovered that patients who had previously received systemic therapy for advanced disease were less likely to develop cirAES.

Published work has estimated the incidence of cirAES from ICIs to range from 30% to 45% for patients receiving ICIs, significantly higher than our work [18]. CTLA‐4 inhibition is generally associated with a higher incidence of cirAES and more severe reactions compared to anti‐PD‐L1/PD‐1 monotherapy [22]. The lower incidence of cirAES in this study may be due to the lower number of patients receiving CTLA‐4 inhibition (which is typically associated with a high incidence of cirAES). Additionally, our retrospective study design may underestimate cirAES compared to a prospectively collated database. Of note, pruritus was the most frequently observed cutaneous irAE, affecting 30% of patients (16/53). This was equally subdivided into cases with an associated maculopapular eruption (15%, 8/53) and isolated pruritus without visible rash (15%, 8/53), consistent with prior literature identifying pruritus as the dominant dermatologic toxicity of immune checkpoint inhibition [16]. Eczematous eruptions were the next most common (26%, 14/53), followed by psoriasiform reactions (23%, 12/53), both representing well‐recognized inflammatory phenotypes associated with ICI therapy a biologic therapy. In managing cirAEs, we identified a low incidence of oral corticosteroids (10%), with most cirAEs grade 2 (79%). Topical corticosteroids may be an appropriate treatment for grade 2 cirAEs, depending on the specific diagnosis. Detailed phenotype‐dependent management guidelines are available from the EADV task force, which are helpful for clinicians in ensuring appropriate management [16]. In a novel finding, we identified that the incidence of cirAES was significantly higher in those who had not received any prior systemic therapy (46% vs. 26%). The use of ICIs in the second‐line setting is higher than in other published works, which may be contributing to the lower incidence of cirAES observed in our study [18]. This important clinical observation may merit further investigation to determine if this is a biological phenomenon related to the immunological impact of prior systemic therapies or if this is related to differences in patient populations who have received prior therapies. Our study's range and spectrum of diagnoses were consistent with the published literature, with pruritus, psoriasis, and pruritus with maculopapular rash being the most common diagnoses [23].

There is significant clinical evidence that the development of cirAEs is associated with improved survival for patients with advanced cancer. In a comprehensive systematic review and meta‐analysis of 30 studies, the authors identified that the development of any cirAE was associated with improved survival [24]. When specifically focused on cirAEs, they identified that the development of a cirAE was strongly associated with an improvement in OS, with an HR of 0.45 [95% CI 0.35–0.59]. This is consistent with our data, which demonstrates a doubling of the median OS for patients who developed cirAEs compared with those who did not. In a large retrospective study of 871 patients with stage III/IV advanced cancer who received ICIs, the authors identified that the development of cirAEs was associated with an improvement in both PFS and OS [25]. In this study, the authors identified a median overall survival for those who experienced mild skin toxicity of 35 months. In another retrospective study of 3731 participants, the authors identified the median OS of approximately 1 year (320 days). Thus, while the survival in our cohort is certainly reassuring, comparison between different studies should be considered with caution, given our small sample size and inclusion of different tumor types. They also performed an adjusted model that considered cirAEs a time‐dependent variable. In this model, they identified that improved PFS and OS persisted for patients who experienced a mild cirAE. A severe cirAE was defined as one requiring oral corticosteroids. In the adjusted model, only OS (not PFS) remained significantly improved for patients who experienced severe cirAEs. In our study, we had only a small number of patients who experienced a severe cirAE (n = 5); thus, further subgroup analyses were not feasible. However, these data were reassuring that even with an adjusted model, the PFS and OS improvements for those who experience mild cirAEs persisted. Thus, our data is consistent with the published literature, affirming a clear association between clinical benefit and development of cirAEs for patients with advanced cancers being treated with ICIs.

This finding raises several important questions for practicing oncologists and dermatologists with respect to balancing the benefits of ICI treatment and toxicity. Do the results of this study and others suggest clinicians should continue ICI treatment beyond toxicity? Can we safely hold ICI treatment in these patients to minimize further toxicity? For dermatologists managing cirAES, do these observations impact our advice whether to continue or stop ICIs? To define the relationship between cirAES and efficacy outcomes from ICIs, there is a need for multi‐institutional collaboration on the collection of prospectively annotated irAE data to enable us to gain a deeper understanding of the complex relationship between irAEs and outcomes. The EADV task force (Dermatology for Cancer Patients) has developed such an essential collaboration, ultimately improving our patients' outcomes. Other studies suggest that for selected irAEs, the grade of toxicity may factor into the decision of balancing toxicity and outcomes [26, 27]. Blood‐based biomarkers of cirAES are under exploration, and it has been suggested that HLA‐DRB1*11:01 and HLA‐DRB1*01:01 BP108 are associated with an increased risk of development of cirAES [28]. Future directions may also investigate whether specific biomarkers of toxicity are associated with PFS or OS from ICIs.

Limitations of our work include the retrospective collection of irAE and clinical data and the analysis that did not account for immortal time bias. These limited detailed assessments, including the exact date of onset, severity, and duration of irAE, may have provided more descriptive data that could be considered in a multivariate analysis. Our single‐institution database would be enhanced by the collaboration of multiple institutions to validate our findings. Although there were some imbalances in study groups (e.g., more patients with melanoma in the cirAES group), our multivariate analysis indicated that the improved PFS and OS observed were independent of age, sex, disease, and prior systemic anti‐cancer therapy.

Future developments in this field may involve the prospective, multi‐institutional collection of cirAE data to further decipher the relationship between cirAE development and efficacy outcomes. Ideally, we would be able to develop a model that may incorporate clinical and other biomarkers, which would risk‐stratify for cirAEs in tandem with clinical benefit to tailor our decisions for individual patients.

5. Conclusions

Our work suggests that cirAES are associated with an improved PFS and OS for patients with advanced cancers independent of age, sex, cancer diagnosis, and prior systemic therapy. For dermatologists, the management of cirAEs is becoming an increasingly significant part of clinical workload, and collaboration with colleagues in medical oncology and pathology to decipher the clinical implications of cirAEs will serve to improve outcomes for our patients.

Conflicts of Interest

David O'Reilly reports conference attendance supported by Pfizer, MSD, and Servier. Muireann Roche reports conference attendance support from UCB, Abbvie, and Janssen. Jarushka Naidoo reports research funding, consulting, advisory board participation, data safety monitoring board service, and/or honoraria from multiple companies, including AstraZeneca (research funding, consulting/advisory board, data safety monitoring board, honoraria), Bristol Myers Squibb (research funding, consulting/advisory board, honoraria), Roche/Genentech (research funding, consulting/advisory board, honoraria), Amgen (research funding, consulting/advisory board), Mirati (research funding), and Daiichi Sankyo (consulting/advisory board, data safety monitoring board, honoraria). She has also served on consulting/advisory boards for NGM Pharmaceuticals, Takeda, Pfizer, Elevation Oncology, AbbVie, and Kaleido Biosciences. Orla Fitzpatrick reports conference attendance supported by AstraZeneca and Pfizer. The remaining authors declare no conflicts of interest.