Imaging and biomarkers to aid in treatment decisions in melanoma and rectal cancer
Bisschop, Kees
DOI:
10.33612/diss.157532721
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Bisschop, K. (2021). Imaging and biomarkers to aid in treatment decisions in melanoma and rectal cancer. University of Groningen. https://doi.org/10.33612/diss.157532721
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
5
Association between pembrolizumab
related adverse events and treatment
outcome in advanced melanoma: Results
from the Dutch expanded access program
Bisschop C.1, Wind T.T.1, Blank C.U.2, Koornstra R.H.T.3, Kapiteijn E.4, Van den Eertwegh A.J.M.5, De Groot J.W.B.6, Jalving M.1, Hospers G.A.P.1
1 Department of Medical Oncology, University Medical Center Groningen, Groningen, The Netherlands.
2 Department of Medical Oncology, Netherlands Cancer Institute NKI-AVL, Amsterdam, The Netherlands.
3 Department of Medical Oncology, Radboud University Medical Center, Nijmegen, The Netherlands.
4 Department of Medical Oncology, Leiden University Medical Center, Leiden, The Netherlands.
5 Department of Medical Oncology, Cancer Center Amsterdam, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
6 Oncology Center Isala, Zwolle, The Netherlands.
Abstract
Toxicity of immune checkpoint inhibitors such as ipilimumab and nivolumab is likely associated with clinical efficacy. In this study, we aim to evaluate this association for pembrolizumab. To this end, data of 147 patients included in the Dutch cohort of the pembrolizumab expanded access program was collected. All data was collected prospectively. Patients with adverse events (AEs) at any time during therapy showed a higher chance of achieving disease control compared to patients without AEs (low-grade AEs vs. no AEs: OR 12.8, p-value = 0.0002, high-(low-grade AEs vs. no AEs OR 38.5,
p-value = 0.0001) according to a multivariate logistic regression analysis. Additionally,
Cox-Regression analysis showed a lower risk of death (HR: 0.51, 95%-CI: 0.27–0.97) and disease progression (HR 0.55, 95%-CI 0.31–0.98) over time for patients with high-grade AEs at any time during therapy compared to patients without AEs during therapy. To correct for time-dependency of occurrence of AEs a pseudo-landmark analysis at 6 months of therapy was performed. Although significance was lost (Wald test p-value > 0.05), a prolonged survival in three patients that stopped therapy within six months due to the occurrence of AEs was observed, suggesting potential treatment benefit despite premature ending of therapy. The occurrence of high-grade toxicity at any time during treatment was associated with higher objective response rates, progression free survival and overall survival. There remains a need to assess the predictive value of early occurring AEs on patient survival.
5
Introduction
The introduction of immune checkpoint inhibitors (ICIs) in advanced melanoma has revolutionized the treatment landscape for patients with advanced melanoma. PD-1 and its ligand PD-L1 are key inhibitory immune checkpoints and Interaction between these molecules results in peripheral T cell exhaustion. This interaction can be disrupted by anti-PD-1 monoclonal antibodies thereby inducing an anti-cancer immune response.1 Impressive clinical benefit has been demonstrated in numerous cancer types and is especially pronounced in patients with advanced melanoma. PD-1 is also involved in maintaining peripheral immune tolerance.2 Therefore, when PD-1 is inhibited, auto-immunity can develop leading to a distinct set of adverse events, termed immune-related adverse events (irAEs).3 IrAEs can be organ specific such as colitis, hepatitis, pneumonitis and hypothyroidism, as well as the result of general immune activation such as fatigue, rash and musculoskeletal problems.4 Up to 20% of patients treated with PD-1 inhibitors experience high-grade adverse events (AEs), these and the more frequent mild toxicity can cause significant morbidity.
Autoimmune toxicity may be sign of a generalized immune response and indeed appears to be positively associated with clinical response. This association was first demonstrated for ipilimumab. In a study of 56 melanoma patients treated with ipilimumab, 36% of patients that developed CTCAE grade 3-4 toxicity experienced a clinical response compared to 5% of patients without high-grade toxicity.5 A combined analysis of two trials (n = 139) confirmed this positive relationship in ipilimumab treated patients.6 In addition, significantly longer median duration of response (35 vs. 18 months) and better overall survival were seen in patients experiencing grade 3-4 AEs. Two larger retrospective analyses (n = 833 and n = 298) of melanoma patients receiving ipilimumab outside clinical trials failed to confirm these associations, potentially due to less accurate collection and grading of adverse events.7,8 A pooled analysis of safety data obtained in four clinical trials with nivolumab in melanoma patients showed that ORR was better in patients with any-grade treatment-related AEs (48.6% vs. 17.8%, P < 0.001).9 Another pooled analysis showed that only cutaneous irAEs rash and vitiligo, were associated with improved survival in melanoma patients treated with nivolumab (P = 0.004 and P = 0.028, respectively).10 The association between toxicity and efficacy has also been shown for other cancer types treated with immune checkpoint inhibitors and for other types of immunotherapy.11-15
Although the efficacy and toxicity of nivolumab and pembrolizumab are comparable, the association between toxicity and efficacy has not previously been determined for pembrolizumab. We aimed to determine whether there was a relationship between toxicity and treatment outcome after correction for important covariables in the Dutch
expanded access program (EAP) pembrolizumab patient cohort. This is the first study to evaluate the association of ICI-related toxicity with treatment outcome in a large group of real-life patients treated with pembrolizumab outside of clinical trials.
Methods
Population
All patients with advanced melanoma treated with pembrolizumab within the Dutch EAP in the six participating centers were included. Patients with uveal melanoma were excluded. The EAP was open between July 2014 and August 2015. Eligible patients had progressed on standard treatment with ipilimumab, and in case of a BRAF-mutated melanoma, a BRAF/MEK inhibitor. Other main inclusion criteria were a minimum age of 12 years and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. Main exclusion criteria were: active central nervous system (CNS) metastases, a history of clinically severe autoimmune disease, a history of life-threatening or severe immune-related AEs on previous treatment with immunotherapy, and treatment with systemic steroids for management of previous irAEs. Patients that were included in the EAP with a baseline performance score of 2 were not excluded from analysis.
Treatment and follow-up
All patients were treated with pembrolizumab 2 mg/kg every 3 weeks for a maximum of two years. Therapy was discontinued in case of disease progression and/or severe toxicity. Routine radiologic assessment consisted of CT of thorax and abdomen and MRI brain in case of brain metastases. This was first performed after 12 weeks of treatment and thereafter according to the treating physicians’ judgment. AEs were treated according to the manufacturer’s treatment algorithms. Clinical data, including data on AEs, were collected prospectively by the treating physician and stored in the patient file.
Data collection
The study was approved by the UMCG local institutional review board (METc 2015/299) and was conducted according to the Dutch guidelines for research involving human subjects. Prospectively collected data was anonymized and collected in a protected electronic database. Data were collected regarding demographics, disease characteristics, treatment characteristics, AEs, treatment of AEs, response to treatment, progression-free survival (PFS) and overall survival (OS). AEs were graded according to the common terminology criteria for adverse events (CTCAE) version 4.0. Response to therapy was assessed using the response evaluation criteria in solid tumors (RECIST) 1.1 criteria.16 Based on the radiological response evaluation, disease control rates (DCR), defined as the presence of at least one complete response, partial response or stable disease during
5
treatment with pembrolizumab, were determined. PFS was defined as the time from start of treatment with pembrolizumab until disease progression (according to RECIST 1.1) or death, whichever occurred first. OS was defined as time from start of treatment until death.
Statistical analysis
The association between DCR and relevant clinical variables was assessed using univariate logistic regression analysis. These variables included: patient gender (categorical), patient age (continuous), ECOG performance status (categorical), AJCC M-stage at start of therapy (categorical), occurrence of CNS metastases at start of therapy (categorical), BRAF mutational status of the tumor (categorical), pre-treatment serum LDH-level (categorical variable differentiating between below or above two times the upper limit of normal), absolute eosinophil count (categorical variable differentiating between below or above the upper limit of normal), absolute leukocyte count (categorical variable differentiating between below or above the upper limit of normal), number of pembrolizumab cycles received by the patient (log-transformed, continuous), number of previous treatment lines received by the patient (continuous), whether or not the patient was previously treated with ipilimumab, BRAF/MEK-inhibition and/or dacarbazine (categorical), whether or not the patient received therapy after pembrolizumab therapy (categorical) and whether or not the patient was treated with corticosteroids during pembrolizumab therapy (categorical). Variables that were univariately associated with DCR were entered in a multivariate logistic regression model for the association between occurrence of adverse events (categorical with no adverse events, low-grade adverse events and high-grade adverse events) and DCR. In a similar manner, the association between each of the variables listed above and OS was assessed in a univariate cox-regression analysis. All variables univariately associated with OS were entered in a multivariate model alongside occurrence of adverse events. The same method was applied for PFS. In the dataset, time at which AEs occurred was not available. To limit the bias caused by time-dependency of occurrence of AEs we performed a pseudo-landmark analysis, for which the landmark time-point was set at six months. This interval was based on previous reports that most AEs occur within the first six months of therapy.17 Survival analysis for OS and PFS in this group was performed similarly as described above. A p-value of ≤ 0.05 was regarded as statistically significant. All statistical analyses were performed using RStudio v1.0.143 for Windows (RStudio, Inc. Boston, MA, USA).
Results
Baseline and treatment characteristics
147 patients were included in this analysis (Table 1). Median follow-up time was 37 months (range 22 - 44 months). Besides 127 patients with cutaneous melanoma, 7 patients with mucosal melanoma and 13 patients with an unknown primary tumor were included. All patients received previous treatment with ipilimumab. In 74 patients (50%) more than one line of systemic treatment had been given before participation in the EAP. The majority of patients had an ECOG performance status of 0 or 1 (85%), stage M1c disease (76%) and a LDH level below two times the upper limit of normal (75%) at the start of treatment. Thirty-nine patients (27%) had known brain metastases at the start of treatment. Patients received a mean of 11 pembrolizumab cycles. Objective responses were seen in 38 of 147 patients (26%) including 8 patients (5%) with a complete response. Thirty-four patients (23%) had stable disease as best response.
Table 1 Baseline and treatment characteristics of patients divided by occurrence and severity of adverse events No adverse events (N = 67) Low-grade adverse events (N = 54) High-grade adverse events (N = 26) Total (N = 147)
Age, median (range) 56 (28 – 86) 59.5 (35 – 80) 56 (31 – 82) 58 (28 – 86) Gender, n (%) Female 30 (44.8) 25 (46.3) 15 (57.7) 70 (47.6) Male 37 (55.2) 29 (53.7) 11 (42.3) 77 (52.4) ECOG-ps, n (%) ECOG-ps 0 28 (41.8) 28 (51.9) 15 (57.7) 71 (48.3) ECOG-ps 1 23 (34.3) 20 (37.0) 11 (42.3) 54 (36.7) ECOG-ps 2 11 (16.4) 3 (5.6) 0 (0.0) 14 (9.5) UNK 5 (7.5) 3 (5.6) 0 (0.0) 8 (5.4)
M-Stage at start of therapy, n (%)
M1a 5 (7.5) 5 (9.3) 3 (11.5) 13 (8.8)
M1b 7 (10.4) 9 (16.7) 3 (11.5) 19 (12.9)
M1c 55 (82.1) 37 (68.5) 19 (73.1) 111 (75.5)
UNK 0 (0.0) 3 (5.6) 1 (3.8) 4 (2.7)
CNS metastases at start of therapy, n (%)
No 44 (65.7) 39 (72.2) 23 (88.5) 106 (72.1)
Yes 23 (34.3) 13 (24.1) 3 (11.5) 39 (26.5)
UNK 0 (0.0) 2 (3.7) 0 (0.0) 2 (1.4)
BRAF-V600 mutational status, n (%)
Wildtype 35 (52.2) 23 (42.6) 10 (38.5) 68 (46.3)
Mutated 26 (38.8) 28 (51.9) 15 (57.7) 69 (46.9)
5
No adverse events (N = 67) Low-grade adverse events (N = 54) High-grade adverse events (N = 26) Total (N = 147) Serum LDH at baseline, n (%) < 2x ULN 48 (71.6) 42 (77.8) 20 (76.9) 110 (74.8) > 2x ULN 12 (17.9) 4 (7.4) 5 (19.2) 21 (14.3) UNK 7 (10.4) 8 (14.8) 1 (3.8) 16 (10.9) Baseline AEC, n (%) Normal 43 (64.2) 38 (70.4) 18 (69.2) 99 (67.3) Elevated 4 (6.0) 7 (13.0) 4 (15.4) 15 (10.2) UNK 20 (29.8) 9 (16.6) 4 (15.4) 33 (22.4) Baseline ALC, n (%) Normal 45 (67.2) 41 (75.9) 22 (84.6) 108 (73.5) Elevated 1 (1.5) 1 (1.9) 0 (0.0) 2 (1.4) UNK 21 (31.3) 12 (22.2) 4 (15.4) 37 (25.2)Number of treatment lines before pembrolizumab, n (%)
1 line 12 (18.0) 11 (20.4) 5 (19.2) 28 (19.0) 2 lines 19 (28.4) 23 (42.6) 11 (42.3) 53 (36.1) 3 or more 8 (11.9) 7 (13.0) 2 (7.7) 17 (11.6) UNK 28 (41.8) 13 (24.1) 8 (30.8) 49 (33.3) Number of pembrolizumab
cycles, median (range) 4 (1 – 35) 10.5 (1 – 38) 12 (1 – 35) 6 (1 – 38) Reason for discontinuation of pembrolizumab, n (%)
Progressive disease 49 (73.1) 27 (50.0) 9 (34.6) 85 (57.8) Death 12 (17.9) 2 (3.7) 0 (0.0) 14 (9.5) Complete response 0 (0.0) 6 (11.1) 1 (3.8) 7 (4.8) Adverse event 0 (0.0) 5 (9.3) 9 (34.6) 14 (9.5) No reduction 3 (4.5) 7 (13.0) 2 (7.7) 12 (8.2) UNK 3 (4.5) 7 (13.0) 5 (19.2) 15 (10.2)
Therapy with corticosteroids, n (%)
No 32 (47.8) 31 (57.4) 8 (30.8) 71 (48.3)
Yes 35 (52.2) 23 (42.6) 18 (69.2) 76 (51.7)
Best overall response, n (%)
Complete response 0 (0.0) 7 (13.0) 1 (3.8) 8 (5.4)
Partial response 5 (7.5) 14 (25.9) 11 (42.3) 30 (20.4)
Stable disease 7 (10.4) 17 (31.5) 10 (38.5) 34 (23.1)
Progressive disease 29 (43.3) 11 (20.4) 2 (7.7) 42 (28.6)
Not evaluated because
of early death 26 (38.8) 5 (9.3) 2 (7.7) 33 (22.4) Overall survival (months), median (95%-CI) 4.6 (1.7 – 7.5) 17.6 (4.2 – 31.0) 28.9 (7.0 – 50.8) 12.7 (7.1 – 18.4) Progression-free survival (months), median (95%-CI) 2.5 (1.9 – 3.1) 10.1 (0.0 – 22.0) 12.0 (8.1 – 15.9) 3.4 (2.1 – 4.7)
Abbreviations: 95%-CI: 95% Confidence interval, AEC: Absolute eosinophil count, ALC: Absolute leukocyte count, CNS: Central nervous system, LDH: Lactate dehydrogenase, SE: Standard error, ULN: Upper limit of normal, UNK: Unknown, ECOG-ps: Eastern Cooperative Oncology Group Performance Status
Toxicity
In total, 67 patients did not develop adverse events, 80 patients developed adverse events of any-grade and 26 patients developed high-grade adverse events (Table 1). Twenty-six patients had AEs that were treated with corticosteroids or other immunomodulating drugs. Most AEs in this cohort were commonly reported pembolizumab-associated AEs (Table 2). In addition, some rare AEs were reported. One patient developed high-grade fever and purpura and was diagnosed with granulomatosis with polyangiitis. This patient was successfully treated with oral cyclofosfamide and intravenous methylprednisolone.18 Another patient developed severe oral mucositis that was treated with clobetasol oral rinse for two months followed by oral prednisolone for two months.
Table 2 Adverse events
Adverse event N = 147
Any grade – no.(%)* Grade 3 or 4 – no.(%)*
Gastrointestinal disorders Colitis Diarrhea 8 (5.4)6 (4.1) 7 (4.8)0 (0.0) Skin disorders Dermatitis Vitiligo Pruritus 9 (6.1) 14 (9.5) 8 (5.4) 1 (0.7) 0 (0.0) 0 (0.0) Endocrine disorders Hypophysitis Hypothyroidism Hyperthyroidism Adrenal insufficiency 2 (1.4) 9 (6.1) 1 (0.7) 1 (0.7) 2 (1.4) 0 (0.0) 0 (0.0) 0 (0.0) Hepatobiliary disorders Hepatitis
Elevated AST and/or ALT Elevated GGT and/or ALP
1 (0.7) 4 (2.7) 7 (4.8) 1 (0.7) 0 (0.0) 1 (0.7) Respiratory disorders Pneumonitis Dyspnea 3 (2.0)2 (1.4) 2 (1.4)1 (0.7) Musculoskeletal disorders Arthritis Arthralgia Eosinophilic fasciitis Myalgia 5 (3.4) 5 (3.4) 1 (0.7) 1 (0.7) 1 (0.7) 0 (0.0) 1 (0.7) 0 (0.0) Other Pancreatitis Elevated amylase/lipase Fatigue Dry mouth Anemia Meningitis Oral mucositis Abdominal pain
Granulomatosis with polyangiitis Other 1 (0.7) 2 (1.4) 18 (12.2) 2 (1.4) 1 (0.7) 1 (0.7) 1 (0.7) 1 (0.7) 1 (0.7) 19 (12.9) 1 (0.7) 2 (1.4) 1 (0.7) 0 (0.0) 1 (0.7) 1 (0.7) 1 (0.7) 1 (0.7) 1 (0.7) 0 (0.0) Total 134 26
5
Association between disease control rate and occurrence of adverse events The occurrence of AEs was univariately associated with DCR (p-value = 1.77x10-6). In addition, baseline serum LDH-level and the number of pembrolizumab cycles received by the patient were significantly associated with DCR by univariate logistic regression analysis (p-values = 0.02 and 2.87x10-9 respectively) (Supplementary Table 1). The median number of pembrolizumab cycles received by patients with versus patients without AEs was comparable (Table 3). A multivariate logistic regression analysis showed patients with high-grade and low-grade adverse events had a higher chance of achieving disease control compared to patients who developed no adverse events (low-grade adverse events vs. no adverse events: Odds ratio (OR) 12.8, p = 0.0002, high-grade adverse events vs. no adverse events: OR 38.5, p = 0.0001; Table 4). Corticosteroid use was not significantly associated with DCR (p-value = 0.69) Eighteen of 26 patients with high-grade adverse events were treated with corticosteroids. DCR did not differ significantly between patients that were or were not treated with corticosteroids in this subgroup (83.3 vs. 87.5%, Fisher’s exact test p-value = 1.0).
Table 3 Disease control rate and number of pembrolizumab cycles received in patients with
and patients without AEs.
No AEs
(n = 67) AEs(n = 80)
Disease control achieved, n (%)
Cycles of pembrolizumab, median (range) 12 (17.9)16 (3 – 35) 60 (75.0)16 (1 – 38)
No disease control achieved, n (%)
Cycles of pembrolizumab, median (range) 55 (82.1)3 (1 – 28) 20 (25.0)4 (1 – 6)
Table 4 Results from the multivariate logistic regression analysis for the association between named variables with disease control rate, defined as stable disease, partial response or complete response.
Variables Odds Ratio LogOdds SE z-value p-value
(Intercept) 164.0 -5.10 1.12 -4.57 4.90e-06
No adverse events 1.00
Low-grade adverse events 12.8 2.55 0.69 3.69 0.0002
High-grade adverse events 38.5 3.65 0.95 3.86 0.0001
Normal serum LDH-levels 1.00
Serum LDH-levels > 2x ULN 3.90 1.36 0.88 1.55 0.1
Number of pembro cycles (cnt.) 1.34 0.29 0.07 4.26 2.06e-05
Abbreviations: cnt.: continuous variable, LDH: lactate dehydrogenase, pembro: pembrolizumab, SE: standard error, ULN: upper limit of normal, ECOG-ps: Eastern Cooperative Oncology Group Performance Status.
Association between survival and occurrence of adverse events
For PFS, univariate Cox Regression analysis showed an association with occurrence of AEs (Wald test p-value = 8.00x10-6), EOCG performance status (Wald test p-value = 2.00x10 -11), baseline serum LDH-level (Wald test p-value = 0.004) and number of pembrolizumab cycles received by the patient (Wald test p-value = 1x10-13) (Supplementary table 2). No association was observed between corticosteroid use and PFS (Wald test p-value = 0.8). Longer PFS was observed in patients who developed high-grade adverse events compared to patients who did not develop adverse events (HR 0.55, 95%-CI 0.31 – 0.98) (Table 5, Figure S1A). For patients who developed low-grade adverse events a trend towards longer PFS compared to patients who did not develop adverse events was observed, however the difference in PFS was not significant (HR 0.63, 95%-CI: 0.39 – 1.01).
Table 5 Results from the multivariate Cox-Regression analysis for the association between
named variables and PFS and OS separately.
Endpoint Variables HR 95%-CI Wald test p-value
PFS No adverse events 1 4.00x10-16
Low-grade adverse events 0.6329 (0.3949 – 1.0143)
High-grade adverse events 0.5479 (0.3069 – 0.9779)
ECOG-ps: 0 1
ECOG-ps: 1 1.9407 (1.2443 - 3.0268)
ECOG-ps: 2 5.1976 (2.4645 – 10.961)
Normal serum LDH-levels 1
Serum LDH-levels > 2x ULN 0.9046 (0.5174 - 1.5815)
Number of pembro cycles (cnt.) 0.9232 (0.8969 - 0.9502)
OS No adverse events 1 2.00x10-15
Low-grade adverse events 0.6634 (0.5075 - 1.0771)
High-grade adverse events 0.5083 (0.2671 - 0.9673)
ECOG-ps: 0 1
ECOG-ps: 1 1.7301 (1.0814 - 2.768)
ECOG-ps: 2 4.2401 (1.983 - 9.0659)
Normal serum LDH-levels 1
Serum LDH-levels > 2x ULN 0.6965 (0.4013 - 1.2086)
Number of pembro cycles (cnt.) 0.9056 (0.8739 - 0.9384)
Abbreviations: cnt.: continuous variable, LDH: lactate dehydrogenase, OS: overall survival, pembro: pembrolizumab, PFS: progression free survival, SE: standard error, ULN: upper limit of normal, ECOG-ps: Eastern Cooperative Oncology Group Performance Status.
Univariate Cox-regression analysis showed an association with OS for occurrence of AEs (Wald test p-value = 2.00x10-5), ECOG performance status (Wald test p-value = 2.00x10-11),
5
baseline serum-LDH levels (Wald test p-value = 4.00x10-4) and number of pembrolizumab cycles received by the patient (Wald test p-value = 5.00x10-14) (Supplementary Table 2). Similar to the observation for PFS, no association was observed between corticosteroid use and OS (Wald test p-value = 0.6). Longer OS was observed in patients who developed high-grade adverse events compared to patients who developed no adverse events in a multivariate Cox Regression analysis including ECOG performance status, baseline serum LDH-level and number of pembrolizumab cycles received by the patient (HR: 0.51, 95%-CI: 0.27 – 0.97) (Table 5, Figure S1B). Although a similar trend was observed for patients with low-grade adverse events compared to patients with no adverse events, there was no significant difference between these groups (HR 0.66, 95%-CI: 0.51 – 1.08).
For the pseudo-landmark analysis 54 patients were available. Based on the univariate Cox-Regression analyses we observed no significant association between the tested variables and OS or PFS (Supplementary Table 3). Of note, no significant difference in OS or PFS was observed with the occurrence of AEs. The Kaplan-Meier curve showed a trend towards shorter OS and PFS for patients who experienced no AEs (Figure S2).
Notably, 5 patients stopped systemic therapy within 6 months due to the occurrence of AEs. After 24 months of follow-up, 3 out of 5 patients were still alive. Although numbers are very small, this proportion is comparable to the 24-month overall survival of patients that did not stop therapy due to adverse events (27 out of 54, 50%).
Discussion
In this study we observed higher DCR in patients with advanced melanoma who developed AEs during pembrolizumab therapy compared to patients who did not. Moreover, we show that in this patient group the occurrence of high-grade toxicity during therapy is associated with longer PFS and OS. This is the first study to show this association in a large cohort of advanced melanoma patients treated with pembrolizumab outside the setting of a clinical trial.
High-grade AEs were reported in 18% of patients on pembrolizumab treatment in the Dutch EAP. In comparison to the Keynote-002 cohort treated with 2mg/kg pembrolizumab, which closely resembles our EAP cohort, overall response rate and median OS are comparable (25.9% vs. 21.0% and 12.7 vs. 13.4 months, for EAP and Keynote-002 cohort respectively; Table 6).19 In comparison to this clinical trial cohort, the rate of high-grade AEs was higher in our EAP cohort (18% vs. 11%). This is most likely the result of a longer follow-up duration in our EAP cohort (median, 37 vs. 10 months). In particular, high-grade colitis was more frequent (4.8% vs. 0%). The rate of high-grade AEs in our EAP cohort is
also high compared to other EAP cohorts (Table 6).20,21 This may be explained by the use of different definitions of AEs. In contrast to other studies, we used all reported AEs in our analysis in order to minimalize investigator interpretation by not differentiating AEs into potentially treatment-related or immune-related nature. Another explanation could be underreporting of AEs in EAP evaluations without prospective data collection focused on AEs.
Table 6 Comparison of efficacy and toxicity data obtained in clinical trials and EAP’s of
melanoma patients receiving pembrolizumab or nivolumab.
Study N ORR (%) Median OS (months) Any-grade AEs (% of patients) High-grade AEs (% of patients)
Dutch EAP 147 25.9 12.7 54.4 17.7
US EAP20 947 14.5 NA 42.2 3.1
Spanish EAP21 67 27 8.7 14.9 1.5
Keynote-00219 180 21 13.4 57 11
Weber et al. (Nivolumab)9 576 31.4 NA 71 10
Freeman-Keller et al. (Nivolumab)10 148 NA NA 68.2* 4.7*
*Only immune-related adverse events were assessed, Abbreviations: AE: adverse event, EAP: Expanded Access Program, ORR: overall response rate, OS: overall survival, US: United States, NA: not assessed
Our data indicates that the occurrence of AEs at any time during therapy is associated with better treatment outcome with pembrolizumab. This is in line with the data previously obtained for nivolumab.9,10 In addition, in this cohort treatment of irAEs with corticosteroid or other immunomodulating treatment did not negatively affect treatment outcome, which was also reported for ipilimumab and nivolumab previously.8,9 Schadendorf et al. showed that early discontinuation of treatment with nivolumab and ipilimumab due to AEs did not affect efficacy outcomes.22 Our data is in line with these findings, although the numbers of patients that discontinue treatment early in our cohort were too small to draw definite conclusions.
This study is limited by its retrospective design. It was not designed to compare efficacy of pembrolizumab between subgroups. Data collection was dependent on reporting of AEs and treatment outcomes by the investigators and data monitoring was not applied. Therefore, non-serious AEs are likely underreported. Subsequent lines of therapy after progression on pembrolizumab were not taken into account and thus survival data may be influenced by patients who rapidly progressed on pembrolizumab or developed serious AEs resulting in a switch to an alternative therapy. The number of patients that discontinue pembrolizumab because of an AE is low (14 of 80 patients with AEs). This can be explained by the fact that some AEs were only lab abnormalities in asymptomatic patients that did not require treatment and treatment decisions were left to the treating
5
physicians in the EAP. The number of individual patients with AEs was too small to describe specific AEs associated with prolonged survival such as previously described by Freeman-Keller et al. for vitiligo.10 Furthermore, the time when a specific AE occurred was not available in the dataset. This may be prone to cause bias towards higher DCR and longer OS and PFS amongst patients developing AEs, because responding patients are exposed to the drug for a longer time period and thus have a longer exposure time in which to develop AEs. Indeed, patients without AEs received significantly less cycles pembrolizumab compared to patients that developed AEs of any kind. However, the multivariate analyses, correcting for the number of cycles of pembrolizumab received by the patients, showed a significant association between occurrence of any-grade AEs with DCR and between the occurrence of high-grade AEs with OS and PFS. Furthermore, we performed a pseudo-landmark analysis to correct for the time-dependency of occurrence of AEs. Although no significant difference between patients with high-grade, low-grade or no AEs was observed in the landmark analysis, it should be noted that the number of patients in this analysis was small. Additionally, early responders who stopped therapy before the six-month landmark may still have experienced treatment benefit. This is illustrated by the 60% 24-month survival rate of the five patients that stopped systemic therapy within 6 months due to the occurrence of AEs. Because of these limitations the pseudo-landmark analysis did not alter the conclusions.
In conclusion, this is the first, large real-life cohort to assess the association of occurrence of AEs and treatment outcome in patients with metastasized melanoma treated with pembrolizumab. The occurrence of high-grade toxicity at any time during treatment was associated with higher objective response rates, progression free survival and overall survival. There remains a need to assess the predictive value of early occurring AEs on patient survival.
Acknowledgements
This work was supported by the Dutch Working Group on Immunotherapy of Oncology (WIN-O).
References
1. Tumeh PC, Harview CL, Yearley JH, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515(7528):568-571.
2. Haanen JB, Thienen H, Blank CU. Toxicity patterns with immunomodulating antibodies and their combinations. Semin Oncol. 2015;42(3):423-428.
3. Naidoo J, Page DB, Li BT, et al. Toxicities of the anti-PD-1 and anti-PD-L1 immune checkpoint antibodies. Ann Oncol. 2015;26(12):2375-2391.
4. Baxi S, Yang A, Gennarelli RL, et al. Immune-related adverse events for anti-PD-1 and anti-PD-L1 drugs: Systematic review and meta-analysis. BMJ. 2018;360:k793.
5. Attia P, Phan GQ, Maker AV, et al. Autoimmunity correlates with tumor regression in patients with metastatic melanoma treated with anti-cytotoxic T-lymphocyte antigen-4. J Clin Oncol. 2005;23(25):6043-6053.
6. Downey SG, Klapper JA, Smith FO, et al. Prognostic factors related to clinical response in patients with metastatic melanoma treated by CTL-associated antigen-4 blockade. Clin Cancer Res. 2007;13(22 Pt 1):6681-6688.
7. Ascierto PA, Simeone E, Sileni VC, et al. Clinical experience with ipilimumab 3 mg/kg: Real-world efficacy and safety data from an expanded access programme cohort. J Transl Med. 2014;12:116-5876-12-116.
8. Horvat TZ, Adel NG, Dang TO, et al. Immune-related adverse events, need for systemic immunosuppression, and effects on survival and time to treatment failure in patients with melanoma treated with ipilimumab at memorial sloan kettering cancer center. J Clin Oncol. 2015;33(28):3193-3198.
9. Weber JS, Hodi FS, Wolchok JD, et al. Safety profile of nivolumab monotherapy: A pooled analysis of patients with advanced melanoma. J Clin Oncol. 2017;35(7):785-792.
10. Freeman-Keller M, Kim Y, Cronin H, et al. Nivolumab in resected and unresectable metastatic melanoma: Characteristics of immune-related adverse events and association with outcomes. Clin Cancer Res. 2016;22(4):886-894.
11. Haratani K, Hayashi H, Chiba Y, et al. Association of immune-related adverse events with nivolumab efficacy in non-small-cell lung cancer. JAMA Oncol. 2018;4(3):374-378.
12. Teraoka S, Fujimoto D, Morimoto T, et al. Early immune-related adverse events and association with outcome in advanced non-small cell lung cancer patients treated with nivolumab: A prospective cohort study. J Thorac Oncol. 2017;12(12):1798-1805.
13. Sato K, Akamatsu H, Murakami E, et al. Correlation between immune-related adverse events and efficacy in non-small cell lung cancer treated with nivolumab. Lung Cancer. 2018;115:71-74. 14. Toi Y, Sugawara S, Kawashima Y, et al. Association of immune-related adverse events with
clinical benefit in patients with advanced non-small-cell lung cancer treated with nivolumab. Oncologist. 2018.
15. Fujii T, Colen RR, Bilen MA, et al. Incidence of immune-related adverse events and its association with treatment outcomes: The MD anderson cancer center experience. Invest New Drugs. 2017. 16. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours:
Revised RECIST guideline (version 1.1). Eur J Cancer. 2009;45(2):228-247.
17. Eigentler TK, Hassel JC, Berking C, et al. Diagnosis, monitoring and management of immune-related adverse drug reactions of anti-PD-1 antibody therapy. Cancer Treat Rev. 2016;45:7-18. 18. van den Brom RR, Abdulahad WH, Rutgers A, et al. Rapid granulomatosis with polyangiitis
induced by immune checkpoint inhibition. Rheumatology (Oxford). 2016;55(6):1143-1145. 19. Ribas A, Puzanov I, Dummer R, et al. Pembrolizumab versus investigator-choice chemotherapy
for ipilimumab-refractory melanoma (KEYNOTE-002): A randomised, controlled, phase 2 trial. Lancet Oncol. 2015;16(8):908-918.
20. Gangadhar TC, Hwu WJ, Postow MA, et al. Efficacy and safety of pembrolizumab in patients enrolled in KEYNOTE-030 in the united states: An expanded access program. J Immunother. 2017;40(9):334-340.
5
21. Gonzalez-Cao M, Arance A, Piulats JM, et al. Pembrolizumab for advanced melanoma: Experience from the spanish expanded access program. Clin Transl Oncol. 2017;19(6):761-768.
22. Schadendorf D, Wolchok JD, Hodi FS, et al. Efficacy and safety outcomes in patients with advanced melanoma who discontinued treatment with nivolumab and ipilimumab because of adverse events: A pooled analysis of randomized phase II and III trials. J Clin Oncol. 2017;35(34):3807-3814.
Supplemental Tables and Figures
Supplementary Table S1 Results from univariate logistic regression analysis for DCR
Variables LogOdds Std.Error z-value p-value
Gender (Intercept) -0,1144 0,2394 -0,478 0,633
male 0,1404 0,3306 0,425 0,671
Age at start (Intercept) -0,013288 0,876519 -0,015 0,988
Cont -0,0004729 0,0147853 -0,032 0,974
ECOG PS (Intercept) 0,429 0,2428 1,767 0,0773
1 -0,429 0,3648 -1,176 0,2395
2 -17,9951 1057,3337 -0,017 0,9864
M-stage at start (Intercept) -0,1625 0,1905 -0,853 0,3935
M1b 1,3665 0,6853 1,994 0,0461
M1c -0,1559 0,5022 -0,311 0,7562
CNS metastases at start (Intercept) 0,07551 0,1944 0,388 0,698
yes -0,54551 0,38226 -1,427 0,154
BRAF status (Intercept) 0,08701 0,241 0,361 0,718
wildtype -0,20479 0,34221 -0,598 0,55
LDH 2xULN (Intercept) -1,1632 0,5123 -2,27 0,02319
normal 1,4191 0,5472 2,593 0,00951
AEC ULN (Intercept) 0,4055 0,527 0,769 0,442
normal -0,2638 0,5643 -0,468 0,64
ALC ULN (Intercept) -3,41E-14 1,41E+00 0 1
normal 1,86E-01 1,43E+00 0,13 0,896
Number of pembro cycles (Intercept) -2,31926 0,3905 -5,939 2,87E-09
Cont 0,27716 0,04974 5,572 2,52E-08
Number of previous treamtent (Intercept) 0,02247 0,21201 0,106 0,916
2 0,02198 0,3659 0,06 0,952
3ormore -0,93876 0,8631 -1,088 0,277
Previous treatment ipilimumab (Intercept) 0,1823 0,4282 0,426 0,67
yes -0,2328 0,4658 -0,5 0,617
Previous treatment BRAF/MEKi (Intercept) 0,06062 0,2011 0,301 0,763
yes -0,25168 0,36953 -0,681 0,496
Previous treatment DTIC (Intercept) -0,01681 0,18335 -0,092 0,927
yes 0,01681 0,46415 0,036 0,971
Treatment after pembro (Intercept) -0,31929 0,47469 -0,673 0,501
Cont 0,04886 0,0744 0,657 0,511
Corticosteroids used (Intercept) 0,02817 0,23738 0,119 0,906
5
Su ppl emen tary T abl e S2 Resul ts from the uni var ia te Co x-regressi on an al yses for OS and PFS O ve ral l s ur viv al Pro gre ssi on fre e s ur vi va l Va ri ab le HR exp (-co ef ) lo w er. 95 upp er .9 5 W al d te st p -v alu e HR exp (-co ef ) lo w er. 95 upp er .9 5 W al d te st p -v alu e Ag e a t s ta rt 1, 00 7 0, 992 9 0, 98 97 1, 02 5 0, 4 1, 00 5 0, 99 51 0,9 88 4 1, 02 2 0, 6 G en de r = m ale 1, 04 4 0, 95 74 0, 70 68 1, 54 3 0, 8 1, 03 9 0,9 62 5 0, 71 81 1, 50 3 0, 8 EC O G P S1 2, 10 7 0, 47 46 8 1, 36 3 3, 257 2, 00 E-11 1, 99 0, 50 25 6 1, 324 2, 991 2, 00 E-11 ECO G P S2 11 ,19 6 0, 08 932 5, 68 22, 06 7 10 ,7 4 0, 093 08 5, 5 20 ,9 87 M -s ta ge a t s ta rt = M 1b 0, 35 77 2, 79 54 0, 14 47 0, 88 46 0, 05 0, 514 8 1, 94 23 0, 24 89 1, 06 5 0, 09 M s ta ge a t s ta rt = M 1c 1, 24 8 0, 80 13 0, 727 7 2, 14 05 1, 213 0, 824 4 0, 70 88 2, 076 CN S m et as tas es a t s ta rt = y es 1, 431 0, 69 87 0, 93 67 2, 18 7 0, 1 1, 50 1 0, 666 1 1 2, 25 4 0, 05 M et as ta tic s ite s a t s ta rt 1, 39 6 0, 716 6 0, 19 34 10 ,0 7 0,7 0, 74 57 1, 34 1 0, 18 24 3, 04 8 0,7 BR AF s ta tus = w ild ty pe 1,2 84 0, 77 91 0, 855 4 1, 92 6 0, 2 1, 26 3 0, 79 18 0, 861 1, 85 2 0, 2 LD H < 2x U LN 0, 39 69 2, 52 0, 237 5 0, 66 31 4, 00 E-04 0, 47 44 2, 10 8 0, 28 53 0, 78 89 0, 00 4 AL C < U LN 0, 397 2 2, 51 8 0, 09 65 8 1, 63 3 0, 2 0, 40 41 2, 47 5 0,0 98 3 1, 661 2, 00 E-01 AE C < U LN 1, 56 7 0, 63 82 0, 75 04 3, 27 2 0, 2 1, 76 9 0, 56 52 0, 88 33 3, 54 4 0, 1 N um be r o f p em br o c ycl es 0, 86 88 1,15 1 0, 837 5 0, 90 12 5, 00 E-1 4 0, 89 77 1,11 4 0, 87 25 0, 92 36 1, 00 E-13 Pr ev io us t re at m en t = 2 1, 37 0,7 29 9 0, 893 6 2, 1 0, 3 1, 39 0, 719 2 0, 91 87 2, 10 5 0, 2 Pr ev io us t re at m en t ≥ 3 1, 53 8 0, 65 01 0, 61 67 3, 837 1, 78 1 0, 561 6 0, 76 94 4, 12 2 Pr ev io us i pi lim um ab = y es 1,15 5 0, 86 59 0, 675 4 1, 97 5 0, 6 1, 09 4 0, 914 0, 666 5 1, 79 6 0,7 Pr ev io us B RAF /M EK i = y es 1, 20 1 0, 8324 0, 777 1, 85 8 0, 4 1, 33 0, 75 19 0, 87 91 2, 012 0, 2 Pr ev io us D TI C = y es 1, 48 8 0, 67 2 0, 89 12 2, 485 0, 1 1, 50 9 0, 662 5 0, 91 61 2, 487 0, 1 Tr ea tm en t a ft er p em br o = y es 0, 97 89 1, 02 2 0,9 03 8 1, 06 0, 6 0,9 96 4 1, 00 4 0, 92 12 1, 07 8 0,9 Co rti co st er oi ds us ed = y es 1,12 2 0, 89 15 0, 75 96 1, 65 6 0, 6 1, 04 1 0, 961 0, 72 01 1, 50 4 0, 8Supplementary Table S3 Results from the univariate Cox regression analyses for the association of covariates with OS and PFS in a pseudo-landmark analysis (N = 54).
Variable Endpoint exp(coef) exp(-coef) lower 95%-CI upper 95%-CI
Age at start OS 1,043 0,9585 0,9962 1,092
Gender = male OS 1,792 0,558 0,6722 4,777
ECOG PS1 OS 1,28 0,7813 0,4958 3,304
M-stage at start = M1b OS 0,4525 2,2098 0,05904 3,469
M stage at start = M1c OS 2,0127 0,4968 0,56984 7,109
CNS metastases at start = yes OS 1,171 0,8538 0,3853 3,56
BRAF status = wildtype OS 1,932 0,5177 0,748 4,989
LDH <2xULN OS 0,3086 3,241 0,07029 1,355
AEC <ULN OS 1,009 0,9914 0,2233 4,556
Low grade AE OS 0,4825 2,072 0,1672 1,392
High grade AE OS 0,397 2,519 0,1118 1,41
Log (number of pembro cycles) OS 0,7608 1,314 0,4566 1,268
Previous treatment = 2 OS 8,98E-01 1,11E+00 0,3194 2,523
Previous treatment ≥ 3 OS 3,77E-08 2,65E+07 0 Inf
Previous ipilimumab = yes OS 0,4075 2,454 0,1523 1,09
Previous BRAF/MEKi = yes OS 0,3601 2,777 0,08269 1,569
Previous DTIC = yes OS 1,823 0,5484 0,5994 5,547
Treatment after pembro = yes OS 0,9048 1,105 0,7418 1,104
Corticosteroids used = yes OS 1,141 0,8761 0,4502 2,894
Age at start PFS 1,022 0,9785 0,9838 1,062
Gender = male PFS 1,899 0,5267 0,8392 4,295
ECOG PS1 PFS 1,328 0,7531 0,6136 2,873
M-stage at start = M1b PFS 0,6629 1,5086 0,1534 2,864
M stage at start = M1c PFS 1,524 0,6561 0,4465 5,202
CNS metastases at start = yes PFS 1,694 0,5905 0,7019 4,086
BRAF status = wildtype PFS 0,4372 2,287 0,05701 3,353
LDH <2xULN PFS 1,521 0,6573 0,6867 3,37
AEC <ULN PFS 0,4448 2,248 0,1035 1,912
Low grade AE PFS 1,6 0,625 0,3693 6,933
High grade AE PFS 0,4208 2,377 0,1666 1,062
Log (number of pembro cycles) PFS 0,4205 2,378 0,1512 1,17
Previous treatment = 2 PFS 0,758 1,319 0,4859 1,182
Previous treatment ≥ 3 PFS 0,8805 1,1357 0,3458 2,242
Previous ipilimumab = yes PFS 1,1951 0,8368 0,158 9,037
Previous BRAF/MEKi = yes PFS 0,5431 1,841 0,2339 1,261
Previous DTIC = yes PFS 0,7872 1,27 0,2932 2,114
Treatment after pembro = yes PFS 1,919 0,5211 0,7069 5,21
Corticosteroids used = yes PFS 0,9206 1,086 0,763 1,111
5
Figure S1 Kaplan Meier curves for (A) progression free survival (PFS) and (B) overall survival (OS) of patients with advanced melanoma treated with pembrolizumab without adverse events (AEs), patients who developed low-grade AEs (CTCAE grade 1 or 2) and patients who developed high-grade AEs (CTCAE grade 3 or 4). Next to the number at risk table the results of the multivariate Cox-Regression analysis with covariates ECOG-ps, baseline serum LDH and number of pembrolizumab cycles received are shown. A significant difference in PFS (HR: 0.55, 95%-CI: 0.31 – 0.98) and OS (HR: 0.51, 95%-CI: 0.27 – 0.97) is shown between patients without AEs and patients with high-grade AEs. Although a trend towards longer PFS and OS was observed for patients with low-grade AEs compared to patients without adverse events, no significant difference was observed. Abbreviations: 95%-CI: 95% Confidence interval, AE: Adverse events, HR: Hazard ratio. A BFigure S2 Kaplan Meier curves for (A) overall survival (OS) and (B) progression free survival (PFS) of patients with advanced melanoma treated with pembrolizumab that survived for at least six months after therapy initiation (6 month pseudo-landmark analysis). Patients are stratified in three groups: patients without adverse events (AEs; red line), patients who developed low-grade AEs (CTCAE grade 1 or 2; blue line) and patients who developed high-grade AEs (CTCAE 3 or 4; green line).
A
