Beyond the Five-Year Milestone: Long-term Survivorship of Melanoma Patients Treated Off-Trial with anti-PD-1

Abstract

Background:

Little is known about the long-term outcomes of anti-PD-1 treated patients with melanoma beyond 5 years, especially for patients treated off clinical trials.

Methods:

This retrospective cohort study includes patients with unresectable stage III/IV non-uveal melanoma treated with anti-PD-1 off trial at Memorial Sloan Kettering Cancer Center between 2014–2017 who survived at least 5 years following their first anti-PD-1 dose (N=139). We characterized overall survival (OS), melanoma-specific survival (MSS) estimates, treatment-free survival rates, and subsequent treatment courses.

Results:

Median follow-up among 5-plus year survivors (N= 125) was 78.4 months (range 60.0–96.3). OS at year 7 (2-years post 5-year landmark) was 90.1% (95% CI: 83.0–94.3%). 14 deaths occurred, 7 due to melanoma. MSS at year 7 (2-yr post 5-yr landmark) was 95.0% (95% CI: 33.5%−95.2%). In patients who completed anti-PD-1 based therapy and did not require subsequent treatment by 5 years (N=80), the probability of not requiring additional treatment for an additional two years was 95.7% (95% CI: 91.0–100%).

Conclusions:

Patients treated with anti-PD-1 regimens off clinical trials who survive at least 5 years from initial anti-PD-1 treatment can be reassured of their excellent long-term prognosis, particularly if they did not require additional melanoma treatment during the first 5 years.

Introduction:

Widespread use of immune checkpoint blockade, particularly regimens containing the anti-programmed death-1 (PD-1) agents nivolumab and pembrolizumab has dramatically improved long-term survival in melanoma.^1,2,3 Follow-up data from clinical trials have now been reported beyond the 5 year milestone. The KEYNOTE-006 trial has reported 7-year follow-up data with median overall survival of 32.7 months in surviving patients treated with pembrolizumab monotherapy.³ The CheckMate 067 trial recently reported data up to 7.5 years with median overall survival of 72.1 months and 36.9 months in the nivolumab plus ipilimumab (nivo + ipi) combination and nivolumab monotherapy groups, respectively.⁴ Since clinical trial data mature before real world standard of care (SOC) data, comparable data beyond 5 years have not been reported for patients treated with anti-PD-1 regimens outside of clinical trials.

We investigated overall survival (OS), melanoma-specific survival (MSS), and treatment-free survival (TFS) as well as subsequent treatment courses of patients who survived at least 5 years following their first dose of anti-PD-1 treatment outside of clinical trial protocols. These results build on recent reports on long-term follow-up of patients treated with immune checkpoint blockades on clinical trials beyond 5-years. Our goal is to provide patients and clinicians with additional information about long-term prognosis for patient populations more likely to reflect standard clinical practice, an increasingly important issue as the number of survivors from advanced melanoma grows.

Materials and Methods:

Study Design

After obtaining institutional review board approval with waiver of informed consent, we conducted a retrospective analysis of all patients with unresectable stage III/ IV non-uveal melanoma treated with anti-PD-1 regimens as standard of care or expanded access program (EAP) and had a minimum of five years documented follow-up at Memorial Sloan Kettering Cancer Center (MSKCC) through July 21, 2022. Tumor responses for best overall response (BOR) were assessed based on clinician subjective determination of response and review of imaging.

Statistical Analysis

This study included patients who had survived at least 5 years from initial anti-PD-1 therapy and had received anti-PD-1 therapy as SOC/EAP. Response to initial anti-PD-1 treatment was estimated with exact 95% confidence intervals. Overall survival was calculated as a landmark 5 years after the start of anti-PD-1 treatment until death. Patients alive at last follow up were censored. Univariable cox regression was used to assess the relationship between baseline characteristics and landmarked OS. Melanoma-specific survival was calculated as a landmark 5 years after the start of anti-PD-1 treatment until death due to melanoma-related causes; deaths due to other causes, and patients alive at last follow up were censored. Treatment-free survival was calculated from start of anti-PD-1 treatment until a patient received additional treatment for melanoma (surgery, radiation, or systemic therapy) or died. Living patients who did not receive additional treatment were censored at last follow up. We also estimated conditional TFS from 5 years after anti-PD-1 treatment start. Kaplan Meier methods were used to estimate these outcomes. TFS was estimated from the time of BOR (complete response [CR], partial response [PR], or stable disease [SD]). Event history plots were used to illustrate overall treatment course and duration. Two-sided p-values less than 0.05 were considered statistically significant. All analyses were performed with SAS 9.4 TS1M6 (The SAS Institute, Cary, NC).

Results:

Patient Characteristics

Between 2014 – 2017, 139 patients received their first anti-PD-1 regimen as standard of care or on expanded access programs and had a minimum of 5 years of documented follow-up through July 21, 2022. Initial anti-PD-1 regimens were: pembrolizumab or nivolumab monotherapy (59.7%, 83/139), or nivolumab plus ipilimumab combination therapy (40.3%, 56/139). Median duration of anti-PD-1 based therapy was 7.1 months (range 0 – 87.5), and 20.1% (28/139) patients stayed on anti-PD-1 therapy for more than 2 years (median 32.2 months, range 24.0–87.5). 51 patients received anti-PD-1 treatment for at least 1 year duration and 28 patients received anti-PD-1 treatment for at least 2 years duration. Patients discontinued anti-PD-1 treatment due to toxicity, disease progression, complete response, death, or maximal clinical response (based on the ASCO Value Framework⁵). Patient and treatment characteristics are shown in Table 1.

Table 1.

Patient, Clinical, and anti-PD-1 Treatment Characteristics

		N (%)
# Patients		139
Age at first anti-PD-1, years	Median (Range) (N=139)	63 (27–88)
Sex	Male	86 (61.9)
	Female	53 (38.1)
Stage at first anti-PD-1	III	24 (17.3)
	M1a	26 (18.7)
	M1b	37 (26.6)
	M1c	34 (24.5)
	M1d	18 (12.9)
Histology	Cutaneous	110 (79.1)
	Mucosal	5 (3.6)
	Acral	3 (2.2)
	Unknown Primary	21 (15.1)
Baseline BRAF Status	Wild type	119 (85.6)
	V600E/K	20 (14.4)
ECOG at first anti-PD-1	0	103 (74.1)
	1	35 (25.2)
	2	1 (0.7)
Pre-anti-PD-1 Liver Mets	Yes	17 (12.2)
	No	122 (87.8)
First anti-PD-1 treatment	Pembro/Nivo Monotherapy	83 (59.7)
	Nivo + Ipi	56 (40.3)

Abbreviations: ECOG, Eastern Cooperative Oncology Group (performance status score). Pembro, pembrolizumab (200 mg IV q3Weeks). Nivo, nivolumab (240 mg IV q2Weeks). Ipi, ipilimumab. Nivo + Ipi (nivolumab 1 mg/kg IV q3Weeks plus ipilimumab 3 mg/kg IV on the same day for maximum of 4 doses; After completing 4 doses of combination therapy, nivolumab 240 mg IV q2Weeks).

Clinical Outcomes after Five Years

Median follow-up time among survivors (N=125) was 78.4 months (range 60.0 – 96.3). From 5 years post-anti-PD-1 treatment start, the median additional follow up in survivors was 18.4 months (range: 0 – 36.3). OS at year 6 (1-year post 5-year landmark) and year 7 (2-years post 5-year landmark) were 92.5% (95% CI: 86.5 – 95.9%) and 90.1% (95% CI: 83.0 – 94.3%), respectively, Figure 1A. Melanoma-specific survival at years 6 and 7 (1 and 2-years post 5-year landmark) were 96.2% (95%CI: 91.2% - 98.4%) and 95.0% (95%CI: 89.1% - 97.8%), respectively, Figure 1B. The MSS of stage III and IV patients in this study was not significantly different (log rank p-value=0.89).

Figure 1:

Overall Survival, Melanoma-Specific Survival, Treatment-Free Survival, and Time to Second Treatment Failure Estimates.

Fig 1A: Overall survival from the landmark of 5-years post initial anti-PD-1 treatment start. 6-year and 7-year OS (1 year and 2 years after the 5-year landmark) were 92.5% and 90.1%, respectively. Tick marks indicate censored patients. Shaded areas represent 95% CI.

Fig 1B: Melanoma-specific survival from the landmark of 5-years post initial anti-PD-1 treatment. MSS at year 6 (1 year post 5-year landmark) was 96.2% and 95.0% at year 7 (2 years post 5-year landmark).

Fig 1C: Treatment-free survival from the time of initial anti-PD-1 treatment. From time of initial anti-PD-1 treatment start, 78.4% (95% CI: 70.6% – 84.4%) and 57.6% (95% CI: 48.9% – 65.3%) patients were treatment failure free at 1 and 5 years, respectively. Median TFS was not reached.

Fig 1D: Time to second treatment failure of those who failed initial anti-PD-1, (N=62). The probability of being free from second treatment failure was 49.1% at 1-year post-initial treatment failure and 33.1% at 5-years post initial treatment failure.

Overall, 14 deaths occurred among this cohort of 5-plus year survivors; 7 died of melanoma. 7 patients had no evidence of melanoma at time of death; 2 deaths were due to other primary malignancies (chronic lymphocytic leukemia and glioblastoma), 3 patients died of other causes, and 2 died of unknown causes. No baseline demographics or pretreatment variables were significantly associated with overall survival, Table S1.

Treatment-Free Survival Following Initial Anti-PD-1

From time of initial anti-PD-1 treatment start, 78.4% (95% CI: 70.6% - 84.4%) and 57.6% (95% CI: 48.9% - 65.3%) patients were treatment failure free at 1 and 5 years, respectively, Figure 1C. Median TFS was not reached. Among patients who discontinued anti-PD-1 treatment and did not require subsequent treatment by 5 years (N=80), the probability of not requiring treatment for an additional 2 years was 95.7% (95%CI: 91.0 – 100%); 3 patients required additional treatment after the 5-year landmark. These 3 patients all had a BOR of CR to initial anti-PD-1. After the 5-year landmark, 2 patients experienced disease recurrence detected by CT scan, 1 developed a new melanoma primary detected from a dermatology follow-up visit. All patients underwent surgery as their subsequent treatment. One patient who experienced disease recurrence was re-challenged with PD-1 monotherapy following surgery, had a complete response to monotherapy, but progressed approximately 3 months after the date of response. This patient subsequently received radiation therapy and is was ongoing treatment, but had no evidence of disease at last follow up. The second patient with disease recurrence had a complete response following surgery and had no evidence of disease as of their last follow up. The third patient with a new primary melanoma had a complete response following surgery and had no evidence of disease as of their last follow up. . OS at year 7 (2-years post 5-year landmark) among these patients who discontinued anti-PD-1 treatment and did not require subsequent treatment by 5 years was 96.1% (95% CI: 88.8 – 98.7%). MSS at year 7 (2-years post 5-year landmark) was 100.0% (95% CI: 100.0–100.0%), 3 deaths, none due to melanoma related causes, occurred after the 5-year landmark among the patients who did not require subsequent treatment by 5 years. Of the 62 patients who required additional therapy after starting their initial anti-PD-1 regimen, the probability of remaining free from second treatment failure was 49.1% (95% CI: 36.1% - 60.9%) at 1-year post-initial treatment failure and 33.1% (95% CI: 21.5% - 45.2%) at 5-years post initial treatment failure, Figure 1D. At time of data cutoff, 2 patients remained on treatment.

Most patients (77.0%, 107/139; 95%CI: 69.1% - 83.7%) experienced a BOR of CR or PR to the initial anti-PD-1 regimen. 82 patients experienced a CR (59%), 25 PR (18%), 17 with SD (12.2%), and 15 with progressive disease (PD, 10.8%). At 1 year, 41.2% (95%CI: 18.6% - 62.6%) of SD patients had not experienced progression or death. When looking at the full follow-up period, all but 1 patient with SD received at least one subsequent treatment (surgery, radiation, or systemic therapy). All patients with disease progression on the initial anti-PD-1 regimen received subsequent treatment. Of the 15 patients whose BOR was PD, 6 received subsequent systemic therapy as follows: anti-PD-1 rechallenge in 1, nivo + ipi in 1, or BRAF ± MEK in 4. The durations of follow up and subsequent treatments following initial anti-PD-1 therapy are detailed in Figure 2.

Figure 2. Treatment Event History.

Black arrows indicate patients with active follow-up. Pink circles represent first progression. Red X’s indicate patient death. Red square indicates patient death due to melanoma-related complications.

Subgroup Analysis of Patients by Treatment Type

Median follow-up among patients treated with anti-PD-1 monotherapy was 78.9 months (range 60.0 – 96.3), and 78.months (range 61.4 – 89.2) for patients treated with nivo + ipi combination. From 5 years post-anti-PD-1 treatment start, the median additional follow up in survivors was 18.9 months (range: 0 – 36.3) and 18.0 months (range: 1.4 – 29.2) among patients treated with anti-PD-1 monotherapy and nivo + ipi combination, respectively. BOR values were not significantly different by treatment type (p=0.37), Table S2. OS at year 7 (2-years post 5-year landmark) was 90.4% (95% CI: 80.6%- 95.4%) in patients treated with anti-PD-1 monotherapy, and 89.8% (95% CI: 76.6% - 95.7%) among patients treated with nivo + ipi combination, Table S3. MSS at year 7 (2-years post 5-year landmark) was 94.9% (95% CI: 87.0% - 98.1%) in patients treated with anti-PD-1 monotherapy, and 95.0% (95% CI: 80.9% - 98.8%) among patients treated with nivo + ipi combination. 79.5% (95% CI: 69.1% - 86.7%) of patients treated with anti-PD-1 monotherapy and 76.8% (95% CI: 63.4% - 85.8%) patients treated with nivo + ipi combination were treatment failure free at 1 year from initial anti-PD-1 treatment. 59.0% (95% CI: 47.7% - 68.7%) of patients treated with anti-PD-1 monotherapy and 55.4% (95% CI: 41.5%−67.2%) patients treated with nivo + ipi were treatment failure free at 5 years from initial anti-PD-1 treatment.

Subgroup Analysis of Patients with Brain Metastases (M1d)

Median follow-up among patients with M1d disease at time of initial anti-PD-1 treatment was 78.0 months (range 66.9 – 95.0). From 5 years post-anti-PD-1 treatment start, the median additional follow up in survivors was 18.0 months (range: 6.9 – 35.0). Among the 18 (13%) patients with M1d disease at time of initial anti-PD-1 treatment and who survived 5-plus years, OS at year 7 (2-years post 5-year landmark) was 94.4% (95% CI: 66.6% - 99.2%). 1 patient died after the 5-year time point (of melanoma complications at 5.41 years). From time of initial anti-PD-1 treatment start, 88.9% (95% CI: 62.4% - 97.1%) and 61.1% (95% CI: 35.5% - 79.2%) of M1d patients were treatment failure free at 1 and 5 years, respectively.

Discussion:

To our knowledge, this study reports the longest follow-up of surviving patients treated with anti-PD-1 regimens outside of clinical trials. This work builds upon the growing body of clinical trial literature reporting long-term follow-up of patients with anti-PD-1 alone or in combination with ipilimumab.^2,5 This real world data is critically important, as clinical trial participants must meet strict eligibility criteria and often have superior clinical outcomes to patients treated in the expanded access and/or approved setting.⁶ Here we demonstrate that melanoma patients treated with anti-PD-1 regimens off clinical trial protocols who survive at least five years have outstanding rates of subsequent overall survival, melanoma-specific survival, and treatment-free survival. Most notably, among patients in this cohort who did not require additional treatment within the first 5 years of starting an anti-PD-1 regimen, patients had a 95.7% chance of not requiring additional melanoma treatment for a subsequent 2 years. Since outcomes 1 and 2 years after our 5-year landmark were generally consistent, we expect that this trend will continue with longer follow up.

These data raise important questions about the role and frequency of surveillance scans for patients who do not require subsequent melanoma therapy within the first 5-years following initial anti-PD-1 treatment. Given that only 3 patients who did not require treatment for melanoma the first 5-years following treatment start subsequently experienced disease recurrence or developed a new melanoma primary after the 5-year landmark, the current practice of semiannual or annual surveillance scans beyond this time point are unlikely to be beneficial. However, these are the first data to demonstrate that surveillance beyond 5 years may not be necessary in this population, and we are cautious about making definitive recommendations given the number of censored patients beyond 5 years and this being only a single-center experience. As the data mature in this study and across larger cohorts, it will be helpful to ensure a similar trend to reinforce these recommendations. It is important to note that a second primary, other skin malignancies, and other cancers do occur following melanoma treatment with anti-PD-1, highlighting the importance of regular dermatology follow-up past the 5-year landmark. Furthermore, given that half of deaths that occurred after the 5-year landmark were from other primary malignancies or alternative non-cancer etiologies, this work highlights the needs for comprehensive survivorship care, including age-appropriate cancer screening and cardiovascular care, for long-term immunotherapy survivors.

Our study has several limitations. Because our goal was specifically to characterize longer-term outcomes among patients who survived at least 5-years post-initial anti-PD-1 treatment to inform about prognosis and survivorship, our results are not generalizable for all patients with advanced melanoma who are initiating anti-PD-1 therapy. Our single center retrospective cohort study reflects a patient population treated at a large, urban, tertiary care cancer center, and additional study in larger cohorts, inclusive of other centers is needed to ensure greater generalizability of our findings.

Overall survival curves for immune checkpoint blockade therapies repeatedly demonstrate long-term plateaus, now including patients treated with anti-PD-1 therapy off clinical trials. Support is building to consider functional cure models.^{7, 8, 9} As the data mature further with additional follow up, and as we add new agents and combinations, such as the recently approved combination of nivolumab plus relatlimab, similar survival and treatment free survival analyses will be invaluable to inform long-term survivorship care.¹⁰

In summary, this report provides the first assessment of long-term survivorship in patients treated with anti-PD-1 therapies outside of clinical trial protocols past the 5-year milestone following the start of initial anti-PD-1 treatment. Our results demonstrate that patients treated with anti-PD-1 therapies and survive at least 5 years after initial anti-PD-1 treatment can be reassured of their excellent long-term prognosis, particularly if they had not required additional melanoma treatment during their first 5 years. This is information is crucial as we begin to plan for long-term survivorship and our experience and knowledge of immune checkpoint blockade and anti-PD-1 treatments grow. And, given the exceptional outcomes of this unique population of long-term survivors following anti-PD-1 treatment for advanced melanoma, these insights serve as promising comparators for other populations of patients with cancer treated with immune checkpoint blockade.

Significance:

Long-term survival of patients with advanced melanoma is now attainable following immune checkpoint blockade with anti-PD-1 regimens. Survival outcomes beyond 5 years are being reported from clinical trials. However, little is known about survival outcomes beyond 5 years for patients treated outside of clinical trials. For the first time, we describe the overall survival, melanoma-specific survival, and treatment-free survival rates beyond five years for patients with advanced melanoma treated with anti-PD-1 therapies off clinical trials. These real world data are key for clinicians to appropriately counsel patients with melanoma about long term prognosis and for institutions to plan survivorship care.