Abstract
Desmoplastic melanoma is a distinct subtype of melanoma known to have preexisting immune infiltrates and high ultraviolet light damage, resulting in a high tumor mutational burden. We hypothesized that this may result in high response rates with single-agent anti-programmed death protein 1 (PD-1) therapy. SWOG S1512 was a two-cohort clinical trial testing the activity of pembrolizumab in patients with surgically resectable (cohort A) and unresectable (cohort B) desmoplastic melanoma. Here we report on the cohort B single-arm clinical trial, which enrolled 27 patients with unresectable desmoplastic melanoma receiving pembrolizumab 200 mg intravenously every 3 weeks for up to 2 years, with the primary endpoint of complete response rate. The complete response rate was 37% (95% confidence interval: 19–58%), and the post hoc endpoint of objective response rate was 89% (95% confidence interval: 71–98%). The estimated secondary endpoints of 3-year melanoma-specific progression-free survival and overall survival were 84% and 96%, respectively, with only one patient having died from melanoma progression. Ten patients (37%) experienced grade 3 or 4 adverse events, and nine patients (33%) discontinued treatment because of adverse events. Patients with advanced desmoplastic melanoma have a high response rate to single-agent PD-1 blockade therapy, supporting single-agent anti-PD-1 as the treatment of choice, but are limited by a frequency of toxicities that is numerically higher than in other patient populations. ClinicalTrials.gov identifier: NCT02775851
Advances in cancer immunotherapy have dramatically improved outcomes for patients with advanced melanoma. Anti-PD-1 as single agent or in combination with anti-cytotoxic T lymphocyte antigen 4 (CTLA-4) or anti-lymphocyte activation gene 3 protein (LAG-3) are standard-of-care frontline treatments for patients with advanced cutaneous melanomas1. Although response rates are higher with combination immune checkpoint blockade therapies, they result in increased toxicities2,3. Because not all subtypes of melanoma have the same etiology, biology and clinical behavior, we postulated that defined melanoma subtypes may have differential responses to immune checkpoint blockade. Desmoplastic melanoma is an uncommon amelanotic subtype of cutaneous melanoma frequently presenting in highly sun-exposed areas of the body and is more common in elderly males4. The histological appearance is that of spindle cells in a background of abundant collagen, with frequent immune cell infiltrates. Immunohistochemically, it is often negative for common melanosomal pathway antigens (Melan-A and HMB45), whereas it is frequently positive for SOX10. Gene sequencing of desmoplastic melanoma biopsies demonstrates a high tumor mutational burden (TMB). Loss-of-function mutations in NF1 are the most common driver oncogenic event, whereas BRAF and NRAS activating mutations are infrequent4–6.
A retrospective case review of over 1,000 patients with advanced melanoma treated with single-agent anti-PD-1/L1 therapy in early clinical trials across 10 academic centers resulted in the identification of 60 patients with advanced desmoplastic melanoma6. This patient cohort demonstrated a best overall response rate of 70% (95% confidence interval: 57–81%) with a median follow-up of 18 months. The complete response (CR) rate was 32%. Responses were seen in patients with desmoplastic melanomas pathologically determined as being of pure and of mixed histological subtypes6. Following these data, we designed SWOG S1512 to test the hypothesis that single-agent anti-PD-1 therapy would have a high response rate both in the neoadjuvant setting (cohort A) and in the surgically unresectable disease (cohort B). The goal of cohort B, presented here, was to prospectively define the CR rate (primary endpoint) of single-agent anti-PD-1 therapy in patients with advanced desmoplastic melanoma who had not previously received systemic therapies. The choice of single-agent anti-PD-1 was based on our anticipation of high response rates in patients with this diagnosis due to the high TMB resulting from high exposure to ultraviolet light carcinogenesis and due to having preexisting immune infiltrates, which are known pathological features of desmoplastic melanoma. Both are tumor characteristics that are associated with objective clinical response to anti-PD-1 therapy7–11.
Results
Study population
Between 6 February 2017 and 17 May 2021, 27 patients were registered at 12 investigator sites in the United States into cohort B of S1512, with clinical and radiological CR rate as the primary endpoint in a single-stage design with 82% power to rule out a CR rate of 5% (the null CR rate) at a one-sided 8.5% alpha level if the true CR rate was 20% (alternative CR rate). Secondary endpoints included progression-free survival (PFS), overall survival (OS) and evaluation of safety and tolerability of pembrolizumab in this setting, and objective response rate (ORR) was a post hoc analysis. The key eligibility criteria were having a desmoplastic melanoma that was deemed to be unresectable and having adequate organ function to receive treatment with pembrolizumab. All patients met the eligibility criteria, received the study therapy as assigned and were assessable for toxicity and response to treatment (Extended Data Fig. 1). The study was permanently closed to accrual on 4 June 2021, as it had reached its accrual goal. As of the final study assessment on 20 May 2024, the median follow-up was 51 months (95% confidence interval: 40–55 months). The median age was 75 years (range, 59–90 years). Ninety-three percent of patients were male, 70% had a performance status of 0 and 30% had a performance status of 1. A normal lactate dehydrogenase (LDH) level was documented in 78% of the patients. The predominant site of the patientsʼ primary desmoplastic melanoma was in the head and neck area (63%), with 19% originating in the extremity and 19% in the torso. Patients enrolled were deemed to have unresectable disease with 30% M0, 4% M1a, 41% M1b and 26% M1c per American Joint Committee on Cancer (AJCC) version 7 (Table 1 and Extended Data Table 1). At the time of final study assessment, all patients were off pembrolizumab treatment. The median number of cycles received was 15 (range, 1–34). Five patients (19%) completed the 2 years of pembrolizumab as planned, and 22 patients (81%) discontinued treatment before the planned duration: nine (33%) due to adverse events and four (15%) due to progression, and nine (33%) were recorded as stopping pembrolizumab due to other, non-protocol specified reasons, including decision by the physician or the patient (n = 6), unrelated sepsis (n = 1), worsening of chronic obstructive pulmonary disease (n = 1) and surgical resection (n = 1) (Extended Data Table 2).
Table 1 |.
Patient and disease characteristics
| Characteristic | (n=27) | |
|---|---|---|
| Age (years) | 75 | (59–90) |
| Sex | ||
| Male | 25 | (93%) |
| Female | 2 | (7%) |
| Race | ||
| White | 27 | (100%) |
| Ethnicity | ||
| Not Hispanic | 26 | (96%) |
| Unknown | 1 | (4%) |
| Performance status | ||
| 0 | 19 | (70%) |
| 1 | 8 | (30%) |
| LDH at baseline | ||
| Elevated LDH | 6 | (22%) |
| Normal LDH | 21 | (78%) |
| Primary site of disease | ||
| Head or neck | 17 | (63%) |
| Extremity | 5 | (19%) |
| Torso | 5 | (19%) |
| AJCC M classification | ||
| M0 | 8 | (30%) |
| M1a | 1 | (4%) |
| M1b | 11 | (41%) |
| M1c | 7 | (26%) |
Patient characteristics among enrolled patients are listed. Median (range) and number (percentage) were reported.
Efficacy
The CR rate was 37% (n = 10, 95% confidence interval: 19–58%). The P value comparing 37% to 5% (the null hypothesis CR rate in the protocol statistical design) was less than 0.001 at a one-sided 8.5% alpha level, which exceeded the 20% target CR rate in the protocol design. The ORR (defined as a CR or partial response (PR) to therapy as per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1) was 89% (n = 24, 95% confidence interval: 71–98%). Responses were frequently deep, with CR or near-CR in most patients (Fig. 1a). They were also rapid, frequently documented in the scans obtained after the first three cycles and durable (Fig. 1b). Responses were noted in visceral and non-visceral metastatic disease sites (Fig. 2). Three patients did not achieve an objective response by radiological assessment. Of these, one patient stopped treatment after five cycles of pembrolizumab to undergo resection and was found to have a pathological complete response (pCR) in the resection sample; however, the radiologic response was stable disease. This patient was excluded from the PFS analysis as the melanoma had been surgically resected. One patient progressed after three cycles of pembrolizumab and then received nivolumab plus ipilimumab outside of this protocol and achieved a clinical CR. One patient withdrew from the study prior to the scheduled restaging imaging and, thus, was considered to not have a response to treatment for the purpose of the analysis. Based on the most recent data lock of 20 May 2024, among the 26 evaluable patients, the PFS at 3 years was 72% (Fig. 3a), and the OS at 3 years was 84% (Fig. 3b). Neither median PFS nor median OS have been reached. There were 10 PFS events, of which four were due to melanoma progression and six were deaths from any cause. Out of nine deaths overall, one was from progressive melanoma and eight were from unrelated causes (Extended Data Table 2). Three cardiovascular events occurred late, after the treatment with pembrolizumab had been discontinued. The estimated 3-year melanoma-specific PFS and OS were 84% and 96%, respectively (Fig. 3c,d).
Fig. 1 |. Depth of response and duration of response.
a, Waterfall plot demonstrating the best response, as percentage change in RECIST target lesions, to pembrolizumab. The vertical bars represent the largest percentage decrease or, if no decrease was observed, the smallest percentage increase in the size of target lesions against baseline. The lines indicate the threshold for objective response (≥30% decrease) or disease progression (>20% increase). One patient did not have follow-up disease assessment data and is denoted on the far left with a gray bar at 50% over baseline. b, Swimmer plot of patient outcomes, including timing of PRs and CRs and PFS events. The horizontal bars represent time from registration. Shaded regions of the bars represent duration of protocol treatment. Patients with arrows are alive and progression free. The patient represented by the gray bar at the bottom withdrew consent after one cycle of treatment; no follow-up outcome data were obtained. nivo-ipi, nivolumab plus ipilimumab; pembro, pembrolizumab.
Fig. 2 |. Examples of responses to therapy.
a, Photographic images of the locally advanced scalp desmoplastic melanoma from a patient, at baseline (left) and on two sequential clinic visits, showing response in cutaneous disease. The baseline lesion had bled, which apparently makes it look pigmented, but the primary lesion was not pigmented. b, PET scan images of the same patient at baseline (left) and at 42 days on-therapy (right) showing response in the cutaneous locally advanced primary (imaged photographically in a) and lymph node metastases. c, Images of CT of the chest of another patient at baseline (left) and at 3 monthsʼ follow-up (right) showing response in lung metastasis. CT, computed tomography.
Fig. 3 |. Analysis of PFS and OS.
a, Kaplan–Meier plot of PFS. The 3-year PFS estimate and 95% confidence interval (CI) are reported. One patient underwent resection and has been excluded from the PFS analysis. b, Kaplan–Meier plot of OS. The 3-year OS estimate and 95% CI are reported. c, Melanoma-specific PFS (MSPFS). The 3-year MSPFS estimate and 95% CI are reported. One patient underwent resection and has been excluded from the analysis. d, Melanoma-specific survival (MSS). The 3-year MSS estimate and 95% CI are reported, with the single event being a patient who died from melanoma brain metastases. YR, year.
Central pathology review
The protocol called for centralized blinded pathology review of the diagnosis of desmoplastic melanoma using archival diagnostic pathology samples of the primary melanoma or a baseline biopsy and the assessment of pathological changes consistent with a response from an on-therapy biopsy. Review was conducted by two independent pathologists (D.A.W. and J.A.P.). Examples of the central pathology review for the desmoplastic melanoma diagnosis are included in Extended Data Fig. 2. Diagnostic pathology samples were not submitted from two cases, and the submitted samples from two additional cases did not contain tumor upon central pathology review (Extended Data Tables 3 and 4). Of the 23 cases with samples evaluable for central pathology review, 19 were confirmed to be diagnostic of desmoplastic melanoma. Of the four cases deemed inconsistent with desmoplastic melanoma based on the provided samples, one case (PT0527) was confirmed to be desmoplastic melanoma in the on-therapy biopsy; two other cases (PT0561 and PT0794) had evidence of a pathological tumor response in their on-therapy biopsies with no tumor content left to confirm desmoplastic melanoma pathology; and one patient (PT0770) had no on-therapy biopsy for analysis. Optional on-therapy biopsies were planned between the first and second infusion of pembrolizumab. Fifteen of the 27 patients had on-therapy biopsies, all of which were evaluable as they were deemed to contain tumor areas upon central pathology review. Examples of the central pathology review of pathological tumor response are included in Extended Data Fig. 3. Four biopsies (27%) had no evidence of pathological features associated with a response to pembrolizumab and were confirmed to be consistent with a desmoplastic melanoma pathologically. Four biopsies (27%) had some regions consistent with a pathological tumor response with presence of residual cancer cells, and seven biopsies (46%) contained evidence of pathological tumor response with no residual cancer cells in the biopsy specimen. In total, 73% of the on-therapy biopsies demonstrated changes consistent with a pathological tumor response (Extended Data Tables 3 and 4).
Toxicity
All 27 patients were assessed for adverse events, and 24 patients (89%) reported treatment-related adverse events of any grade. Ten patients (37%) experienced grade 3 or 4 adverse events, and nine patients (33%) discontinued treatment because of adverse events. Treatment-related toxicities with the highest frequency of all grade events included fatigue (n = 15, 56%), diarrhea (n = 9, 33%), maculopapular rash (n = 8, 30%), pruritus (n = 6, 22%), anemia (n = 5, 19%), arthralgia (n = 5, 19%) and decreased lymphocyte count (n = 5, 19%). Adverse events of interest are shown in Table 2 and Extended Data Tables 5 and 6, and a listing of all adverse events is included in Supplementary Table 1.
Table 2 |.
Number of patients with treatment-related adverse events
| Any grade | Grade 3–5 | |
|---|---|---|
| Any adverse event (AE) | 24 (89%) | 10 (37%) |
| Any serious adverse event (SAE) | 7 (26%) | 5 (19%) |
| AE/toxicity led to discontinuation of treatment | 9 (33%) | - |
| Immune-related AE | 18 (67%) | 8 (30%) |
| SAE | 6 (22%) | 5 (19%) |
| AE led to death | 0 (0%) | 0 (0%) |
All adverse events reported were assessed as possibly, probably or definitely related to the study treatment. Patients included in the safety analysis must have received at least one dose of protocol therapy. Adverse event severity was scored using the National Cancer Institute CTCAE version 4.0. No grade 5 (led to death) adverse events were reported.
Genomic analyses of tissue biopsies
Whole-exome sequencing (WES) analysis was performed in 32 tumor samples (including 15 baseline and 17 on-therapy samples from 20 patients) to assess TMB and canonical genetic driver mutations of melanoma (Extended Data Fig. 4). The dataset was filtered to exclude WES from samples that did not contain tumor, either because the provided sample did not have cancer cell content or because the biopsy had a pathological tumor response without residual viable cancer cells by central pathology review; this excluded three patients from the analysis, leaving a total of 17 patients with biopsies amenable for genetic sequencing analyses. The median TMB across biopsies was 82.1 mutations per megabase (Mut/Mb; range, 16.4–230 Mut/Mb). This is an order of magnitude higher than in prior datasets of cutaneous melanoma, where the reported median TMB was 7.55 Mut/Mb12, and similar to that previously described for desmoplastic melanoma5,6. Tumor biopsies from 14 of 17 patients (82.4%) had NF1 mutations; 10 of these were putative loss-of-function mutations (one splice acceptor, one splice donor and eight nonsense mutations; Extended Data Fig. 4). Recurrent mutations were also found in TP53 (n = 13, 76.5%), ROS1 (n = 11, 64.7%) and KMT2A (n = 10, 58.8%). Notably, biopsies from 10 of 17 patients (58.8%) had mutations in at least three of these four genes. No driver mutations were detected in BRAF or NRAS, the most common canonical oncogene driver mutations of cutaneous melanoma (Supplementary Table 2).
The only patient who had documented progressive disease by clinical objective response assessment (PT0710) while on pembrolizumab treatment subsequently responded to the combination of nivolumab with ipilimumab. In the baseline and on-therapy biopsies (with estimated tumor cellularity of 32% and 66%, respectively), two putative loss-of-function mutations were detected in NF1, including a splice donor variant (at 13% and 30% variant allele frequency (VAF) at baseline and on-therapy, respectively) and a nonsense mutation (R2517*, at 16% and 30% VAF). Two missense mutations were detected in ROS1 (P2130S, at 18% and 29% VAF; G1809R, at 14% and 24% VAF); one missense mutation in TET2 (H880Y, at 13% and 31% VAF); and one missense mutation in JAK1 (D608E, at 13% and 27% VAF). Although these genes were also mutated in biopsies from other patients, the sample from PT0710 also had the canonical gain-of-function mutation in PIK3CA, V344G, in both the baseline and on-therapy biopsies (at 19% and 29% VAF, respectively). No other putative drivers were annotated in PIK3CA in any other patients in this cohort. Two patients had PR as best response on study (PT0523 and PT0770) but later had progression with increase in the size of target lesions. PT0523 had desmoplastic melanoma confirmed in the diagnostic biopsy. The on-therapy biopsy was not collected. The baseline biopsy from PT0770 was not confirmed to contain desmoplastic melanoma by central pathology review, and the on-therapy biopsy was not collected. WES data were not available from biopsies of either of these patients, and there were no available biopsies from the time of progression. One patient (PT0592) had a CR with later progression due to the development of a new scalp lesion. In reviewing this case, this patient had confirmed desmoplastic melanoma histology at baseline and no evidence of features consistent with a pathological response in the on-therapy biopsy. The baseline and on-therapy biopsies (with estimated tumor cellularity of 18% and 27%, respectively) had two mutations in TP53 (hotspot loss-of-function mutation R248W, at 19.9% and 16.6% VAF, respectively; P151F, at 18.6 and 11.4% VAF), a nonsense mutation in CDKN2A (R58*, at 28% and 20% VAF) and missense mutations in ROS1, KMT2A, ALK, FGFR2 and RET. No sampling of the progression lesion was available for WES analysis (Supplementary Table 2).
Discussion
This prospective clinical trial provides evidence that patients with advanced desmoplastic melanoma are exceptional responders to single-agent anti-PD-1 therapy. The cancers with the highest response rates to single-agent anti-PD-1 therapy are Hodgkin lymphoma, microsatellite instability (MSI) colorectal cancer and Merkel cell carcinoma, to which this report adds desmoplastic melanoma11,13–17. These cancers have strong cancer-specific antigens, either because they are driven by viral infections (Epstein–Barr virus for Hodgkin lymphoma and Merkel cell polyoma virus for a subset of Merkel cell carcinoma) or because they have a high mutational load (from DNA mismatch repair deficiency in MSI-high colorectal cancers and from the carcinogenic effect of ultraviolet light exposure in desmoplastic melanoma and for another subset of Merkel cell carcinoma)11,18. The high response rate to anti-PD-1 therapy in patients with desmoplastic melanoma occurred despite several features that have been proposed to be adverse for responses to cancer immunotherapy, such as being a cancer defined by the high presence of fibrosis and collagen (desmoplasia is the pathological description of a fibrous reaction) and being an undifferentiated cancer clinically defined in part by the lack of pigmentation and pathologically by the lack of expression of melanosomal antigens4,19. However, it is a cancer defined pathologically as frequently having immune infiltrates and is among the most highly mutated cancers, with C > T mutation patterns implicating ultraviolet light radiation as the dominant carcinogen5,6. Therefore, the positive features of preexisting immune infiltrates recognizing a highly mutated cancer dominate over fibrosis and dedifferentiation as negative features of response to anti-PD-1 therapy.
Desmoplastic melanoma occurs more commonly in elderly patients as evidenced in this series and others5,6,20. Multiple comorbidities are more prevalent in such a population, and those comorbidities add to potential complications and, thus, limitations for anticancer therapies. Only five patients completed the intended 2-year treatment regimen, with two-thirds of the patients stopping therapy early due to either adverse events or physician or patient decision. The toxicity grades and rate of treatment discontinuation in cohort B of S1512 were higher than the reported toxicities in an analysis of 8,937 patients with different cancers treated with the single agent pembrolizumab21, which would be expected in this population of elderly patients with multiple comorbidities4. Eight of the nine deaths in this series were due to comorbidities reported as unrelated to the desmoplastic melanoma or the treatment with pembrolizumab. It cannot be ruled out that the late cardiovascular and pulmonary events that occurred after discontinuation of the pembrolizumab treatment are completely unrelated to the treatment with anti-PD-1, although the causality assessed by the local investigators was of being from unrelated conditions. The patients in this series entered the clinical trial with different stages of desmoplastic melanoma, some of them being considered not benefitting from further local therapies without having systemic metastases and others with systemic spread, including to visceral metastases. The benefit of systemic therapy with pembrolizumab appears to be independent of disease extent, whether locally advanced or metastatic. Long-term responses were noted both in patients who had a CR to treatment and in patients who had a PR to treatment, highlighting that residual findings in patients with PR may not be due to active melanoma. The use of positron emission tomography (PET) or tumor biopsies could help better assess residual findings. Collectively, these data highlight that the responses to therapy were durable in most patients, and even the few patients who had a relapse were likely to not die of the desmoplastic melanoma or from toxicities from pembrolizumab.
In conclusion, patients with unresectable desmoplastic melanoma have a high response rate to single-agent anti-PD-1 therapy, with responses most frequently being deep and rapid, as noted in the first 9-week scans. Responses were durable, with ongoing responses beyond the discontinuation of pembrolizumab, and most of the patient deaths were due to unrelated causes. Notably, only one patient in this series is known to have died from progressive melanoma. The delivery of the treatment was limited by the toxicities in this patient population with increased comorbidities. Not completing the intended 2 years of treatment with pembrolizumab did not appear to affect the rate or duration of responses. The desmoplastic variant of melanoma should be considered a predictive biomarker of response to single-agent anti-PD-1 blockade. Treatment of choice should be single-agent anti-PD-1 blockade rather than combination immune checkpoint blockade therapy with ipilimumab (anti-CTLA-4) or relatlimab (anti-LAG-3).
Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41591-025-03875-5.
Methods
Participants
Patients were eligible if they were at least 18 years of age with a histologically or cytologically confirmed diagnosis of unresectable desmoplastic melanoma and measurable disease according to RECIST version 1.1. Eligible patients had adequate bone marrow function, adequate liver function and a performance status of 2 or lower. Prior surgery was allowed. Patients with locally advanced, unresectable, primary diagnosis of desmoplastic melanoma or distant metastatic disease were eligible. Main exclusion criteria included prior systemic treatment for the current melanoma, active autoimmune disease requiring systemic treatment within 2 years of study entry, live vaccines within 42 days prior to registration and any history of brain metastasis.
Trial design and treatments
S1512 was a phase 2 clinical trial with two separate cohorts to assess the efficacy of pembrolizumab in patients with desmoplastic melanoma. Cohort A (not described here) was designed to evaluate the pathological CR rate in patients with resectable desmoplastic melanoma treated with neoadjuvant pembrolizumab. Cohort B (described here) evaluated the clinical CR rate in patients with unresectable desmoplastic melanoma treated with pembrolizumab. Pembrolizumab at 200 mg was given intravenously every 3 weeks for a planned period of up to 2 years. Tumor measurements were performed prior to initiation of therapy and every 9 weeks. Patients whose disease became resectable were allowed to proceed with surgical resection. Adverse events were assessed at baseline and every 3 weeks while undergoing treatment. Patients were planned to be followed for PFS and OS for 5 years from the time of registration. Biopsy samples and whole blood were obtained for correlative analysis prior to the first pembrolizumab treatment and at approximately 9 weeks of therapy.
Outcomes
The primary endpoint was the rate of clinical CR, defined as no radiographic or clinical evidence of disease. Disappearance of all target and non-target lesions was required for a CR. Secondary endpoints included PFS, OS and evaluation of safety and tolerability of pembrolizumab in this setting. Disease assessments, radiographically or by clinical examination documented by photos with ruler, occurred at baseline and every 9 weeks after the start of pembrolizumab. RECIST version 1.1 criteria were used for clinical response assessment. OS was measured from registration to the date of death from any cause, with the patients last known to be alive censored at the date of last contact. PFS, estimated only among patients who did not undergo resection, was measured from the date of registration to the date of first documentation of progression, symptomatic deterioration or death due to any cause, with patients last known to be alive and without report of progression censored at the date of last contact. Adverse events were assessed by the investigators using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 4.03.
Central pathology review of desmoplastic melanoma diagnosis
Desmoplastic melanoma was defined by the presence of previously published criteria of spindle tumor cells, mostly with elongated hyperchromatic nuclei and mild to moderate cytologic atypia with abundant stromal collagen (desmoplasia)4,22. Criteria for pure desmoplastic melanoma were met if more than 90% of the tumor contained a low density of tumor cells as described above, imparting a scar-like low-power (20×) histologic pattern. Mixed desmoplastic melanoma contained significant (>10%) areas of tumor with a higher cell density and less stromal collagen.
Sample collection and nucleic acid extraction
Biopsies were individually procured at each study site and formalin fixed and paraffin embedded (FFPE) and then stored as either 4–5-μm slides or blocks using standard clinical pathology practices. Whole blood samples were collected in 10-ml tubes. Collected samples were sent to the University of California, Los Angeles (UCLA) for centralized storage and further processing. Upon arrival, whole blood samples were processed using Ficoll-Hypaque density gradient centrifugation to isolate peripheral blood mononuclear cells (PBMCs), which were stored in liquid nitrogen. DNA was extracted from PBMC samples using a Qiagen AllPrep DNA/RNA Mini Kit (cat. no. 80284), according to the manufacturerʼs protocol.
The Translational Pathology Core Laboratory at UCLA performed sectioning (5 μm) from FFPE blocks, and slides were individually stained for mouse anti-human S100 (Cell Marque, cat. no. 330M, 1:400, for melanoma cells), mouse anti-human CD8 (Agilent Technologies, M7103, 1:100, for cytotoxic T cells) and hematoxylin and eosin (H&E) (for histological assessment). Slides were imaged at ×40 magnification for central pathology review. Sequential cuts or slides were used for nucleic acid extraction (30–50 μm) using a Covaris truXTRAC FFPE tNA Ultra Kit (SKU no. 520252) using either a Covaris M220 or E220 evolution Focused-ultrasonicator platform for the Adaptive Focused Acoustics (AFA)-enhanced paraffin emulsification and tissue rehydration steps according to the manufacturerʼs protocol, with some exceptions. For scrolls that were sectioned from FFPE blocks, each scroll was placed into its own Covaris microTUBE-500 (SKU no. 520185). Each scroll was individually processed for paraffin emulsification, proteinase K treatment, centrifugation and de-crosslinking, until the column binding steps. Then, the suspended RNA or DNA was combined into the same respective column for purification. For samples preserved as slides, each slide was processed one at a time (up to 10 slides, based upon the amount of tissue available) by placement on a slide heater at 40 °C for 40 seconds. Sample was scraped off the individual slides using a Covaris FFPE Tissue Pick or a Covaris FFPE Section Pick, and the material from all slides was combined in an individual Covaris microTUBE-500. If the tissue pick was used, the pick was added to the tube after scraping the final slides. The sample was processed in a single tube for the remainder of the protocol.
WES
DNA was evaluated and quantified on an Agilent TapeStation 4200 using Genomic DNA ScreenTape (cat. no. 5067–5366) reagents. Then, 100 ng of DNA was diluted in 50 μl of nuclease-free water for WES library preparation; for a subset of samples, if there was not 100 ng available, the total amount of DNA was used. Libraries were prepared at Agilent Technologies on the Agilent BRAVO platform using an Agilent SureSelect XT HS2 DNA (cat. no. G9984A) library preparation kit. Exome capture was done using Agilent SureSelect Exome v7 capture probes spiked in with equimolar concentration of Agilent OneSeq Backbone probes (300-kb whole genome-wide resolution). WES libraries were evaluated and quantified on the Agilent TapeStation using the High Sensitivity D1000 Assay. After quantification of post-capture libraries, samples were reviewed and did not proceed with sequencing if they had molarity concentrations less than 1 nM. Libraries that passed quality control were pooled for a final concentration of 2 nM and sequenced on the Illumina NovaSeq platform (S4 flow cell, 2 × 150-bp reads). WES libraries were sequenced for an average of 187× depth in tumor specimens (with an average of 82% of the targeted exome having 100× coverage) and 157× depth in PBMC samples (with an average of 88% of the targeted exome having 50× coverage). Data processing, including sequencing alignment, somatic variant detection and tumor purity estimation, was performed as previously described12. Both histopathological analysis and sequencing quality assessment were used to exclude samples due to low tumor purity (lack of tumor content observed in the biopsy).
Genomic analysis of tumor specimens
Each sample was analyzed for melanoma genetic subtypes (BRAF activating mutations, NRAS hotspot mutations and NF1 mutant or loss-of-function mutations). Non-silent TMB was quantified by normalizing the number of non-silent variants with at least 5% VAF by the size of the genome (in Mb) with at least 50× depth of coverage. Samples were evaluated for both high tumor cellularity (by manual review of immunohistochemistry and S100-stained slides) and adequate sequencing coverage for inclusion in the genomic reporting.
Trial oversight
Approval for the trial was obtained through the Cancer Therapy Evaluation Programʼs Central Institutional Review Board. The trial was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines, and the study protocol was approved by the relevant ethics bodies at each participating site. Patients provided written informed consent before being screened for enrollment.
Statistical analysis
The primary endpoint was clinical CR rate, per RECIST 1.1. The sample size (n = 21) was based on a single-stage design with 82% power to rule out a CR rate of 5% (the null CR rate) at a one-sided 8.5% alpha level if the true CR rate was 20% (alternative CR rate). The observation of three out of 21 CRs would be considered evidence that the treatment warrants further study, provided that other factors, such as toxicity and OS, were also favorable. Enrollment of 26 patients was estimated to allow for at least 21 evaluable patients. Surgical resection was allowed if the tumor was converted from unresectable to resectable. Assessment of CR required clinical disappearance of disease. Thus, a pCR would not be considered a CR. All patients who received a single dose of study treatment and met eligibility criteria were considered evaluable for the primary endpoint. A one-sided exact binomial test using the method of Clopper and Pearson was used to test the clinical CR rate against the null hypothesis of 5%. For binary outcomes, exact confidence intervals were constructed using the method of Clopper and Pearson23. The Kaplan–Meier method was used to estimate the distributions of PFS and OS, and the log–log method was used to estimate the corresponding confidence intervals for survival at 3 years. Melanoma-specific survival was calculated as 1 – CIF, where CIF is the cumulative incidence of melanoma-specific deaths estimated non-parametrically using the method of Nelson–Aalen. Death from other causes were treated as competing risks. Melanoma-specific survival confidence interval for survival at 3 years was calculated using the method recommended by Pintilie12,24,25. Median follow-up was estimated using the reverse Kaplan–Meier approach. All analyses were performed using SAS 9.4 (SAS Institute, Inc.) and R version 4.3.1 (The R Foundation for Statistical Computing) software.
Extended Data
Extended Data Fig. 1 |. CONSORT diagram.
Patient eligibility and disposition. PFS: Progression-free survival.
Extended Data Fig. 2 |. Examples of central pathology review of desmoplastic melanoma diagnosis.
Each biopsy had 2 sequential sections stained and imaged for a) S100 or b) H&E, and images were reviewed simultaneously by pathologists to confirm the presence of desmoplastic melanoma by viable tumor cells, shape, and collagenous stroma. In situations with multiple biopsies, pathologists reviewed all sections in parallel to determine consensus annotation.
Extended Data Fig. 3 |. Examples of central pathology review of response to therapy in baseline and on-study biopsies.
Each biopsy had 2 sequential sections stained for S100 (left) or H&E (right), and images were reviewed simultaneously by pathologists to determine presence of a) viable tumor regions with desmoplastic melanoma histology or b) features indicative of tumor regression and immune-mediated pathological response.
Extended Data Fig. 4 |. Tumor mutational burden and recurrently mutated genes.
Whole exome sequencing data was available from biopsies from 17 patients. Tumor mutational burden (TMB, in Mutations [Mut] per Megabase [Mb], top panel) and genes with mutations in at least 25% of cases are shown (bottom panels). Mutations are annotated by their putative impact (fill color) and whether they are present in chromosomal regions with loss-of-heterozygosity (LOH). Patients are further annotated based upon their clinical response by imaging at 9 weeks (complete response [CR], partial response [PR], stable disease [SD], or progressive disease [PD]).
Extended Data Table 1 |.
Individual patient characteristics and central pathology review
| ID | Age range | Sex | PS | Target Lesions | Nontarget Lesions | AJCC Stage |
|---|---|---|---|---|---|---|
| PT0522 | 80–90 | M | 1 | Lung - right lower lobe, Lung - right perihilar node Center forehead - umbilicated tan papule |
Lung - left lower lobe | T3aN2cM1b |
| PT0523 | 80–90 | M | 0 | Lung - right hilum (lateral), Lung - right hilum (medial), Lung - mediastinum, Lung - parenchyma, |
T4aN0M1b | |
| PT0582 | 70–80 | M | 0 | Lung - left base | T4aN0M1b | |
| PT0535 | 80–90 | M | 0 | Lung - left upper lobe, Scalp - right frontal |
Calvarium | TXNXM1b |
| PT0561 | 80–90 | M | 1 | Lung - right lower lobe, Liver - right lobe, Liver - left lobe, Retroperitoneal - right retrocrural mass, Retroperitoneal - perinephric mass |
Liver - multiple hypoattenuating lesions, Soft tissue - bilateral posterior thigh |
T4aNXM1c |
| PT0565 | 80–90 | M | 0 | Soft tissue - Left posterior ankle | Lymph nodes - left femoral | T2bN1bM0 |
| PT0607 | 60–70 | M | 1 | Lung - right lower lobe, Pectoralis mass - right |
Pleural effusion - left Pleural nodules - left |
T0N0M1c |
| PT0570 | 70–80 | M | 0 | Scalp - vertex, Scalp - right, Hepatic Lobe - left |
Scalp - multiple soft tissue nodules, pulmonary nodules - multiple, bilateral, periauricular lesion, postauricular lesion |
TXNXM1c |
| PT0592 | 80–90 | M | 0 | Superior forehead | T3bN0M0 | |
| PT0527 | >80–90 | M | 1 | Soft tissue - left intramuscular lesion, Soft tissue - posterolateral left muscular lesion |
Soft tissue - left neck | T4bN3M0 |
| PT0710 | 80–90 | M | 0 | Cheek/nasolabial fold - right | T1aN0M0 | |
| PT0650 | 70–80 | M | 0 | Lung - left upper lobe, Lung - left lower lobe, Anterior chest wall - left, Parotid mass - right, Retromastoid mass - right |
Lung - right lower lobe, Lung - right middle lobe, Lung - smaller pulmonary metastases |
TisN1bM1a |
| PT0628 | >50–60 | F | 0 | Lung - left lower lobe | T4bNXM1b | |
| PT0706 | >70–80 | M | 0 | Lung - left upper lobe, Chest wall - anterior |
Lung - multiple sites | T3aN0M1b |
| PT0723 | 60–70 | M | 0 | Lung - left lower lobe, Lung - basilar right lower lobe |
Lung - pulmonary nodules | T4bN2aM1b |
| PT0718 | 60–70 | M | 0 | Forearm - right distal radial | T4aNXM0 | |
| PT0623 | 80–90 | M | 1 | Lung - left upper lobe, Liver - right lobe, |
Cervical lymph node - right lower posterior Lung - left upper lobe, |
T4aN3M1c |
| Liver - left lobe | Lung - right upper lobe, Lung - right lower lobe, Renal - left, Rib - 7th rib, Liver- right lobe |
|||||
| PT0655 | 60–70 | M | 0 | Soft tissue - left preauricular | T4aN2bM1b | |
| PT0681 | 60–70 | M | 0 | Lip - midline lower | T3bN0M0 | |
| PT0772 | 60–70 | M | 1 | Back - left posterior, Liver - segment 4 |
T4aN0M1c | |
| PT0770 | 80–90 | M | 1 | Lung - left pleural, Lymph node - prevascular, Liver |
Lung - left upper lobe | T4aN0M1c |
| PT0777 | 60–70 | M | 0 | Lung - left upper lobe, Lung - right middle lobe |
Lung - right upper lobe, Lung - right |
T2aNXM1b |
| PT0796 | 80–90 | M | 0 | Lymph node, left cervical level 1B, Lymph node - left cervical level 2a/3, Omental nodule |
T0N2bM1b | |
| PT0797 | 70–80 | M | 0 | Lung - left upper lobe | T3aN0M1b | |
| PT0794 | 70–80 | F | 0 | Lung - right upper lobe, Axillary node - right |
Pulmonary nodules, Spinal mass - T3 vertebra |
T4bN0M1c |
| PT0839 | 70–80 | M | 0 | Postauricular - left | T4bNXM0 | |
| PT0789 | 70–80 | M | 1 | Face - right | T3aN0M0 |
F, female; M, male; PS, performance status.
Extended Data Table 2 |.
Individual patient treatment summary and outcomes
| ID | Number of Infusions | Reason for Discontinuation | Best Clinical Objective Response | Follow up / Progression | PFS (Months) | OS (Months) | Status / Cause of Death |
|---|---|---|---|---|---|---|---|
| PT0522 | 10 | Adverse event or side effect | PR | No | 58 | 58 | Alive |
| PT0523 | 7 | Progression | PR | 20% Increase | 7 | 65 | Alive |
| PT0582 | 33 | Treatment complete | CR | No | 49 | 49 | Dead - stroke |
| PT0535 | 34 | Treatment complete | PR | No | 42 | 42 | Dead - sepsis |
| PT0561 | 14 | Adverse event or side effect | PR | No | 13 | 13 | Dead - pulmonary fibrosis |
| PT0565 | 3 | Adverse event or side effect | PR | No | 33 | 33 | Alive |
| PT0607 | 1 | Refusal unrelated to adverse event1 | NA1 | No | 1 | 1 | Alive |
| PT0570 | 20 | Other - patient and physician decision | PR | No | 54 | 54 | Alive |
| PT0592 | 15 | Progression | CR | New scalp lesion | 11 | 51 | Dead - acute myelogenous leukemia |
| PT0527 | 7 | Other - patient and physician decision | CR | No | 33 | 33 | Dead - cardiac related |
| PT0710 | 3 | Progression | PD | 20% Increase | 2 | 44 | Dead - kidney failure |
| PT0650 | 17 | Adverse event or side effect | uCR | No | 60 | 60 | Alive |
| PT0628 | 33 | Treatment complete | CR | No | 60 | 60 | Alive |
| PT0706 | 13 | Other - patient and physician decision | PR | No | 52 | 52 | Alive |
| PT0723 | 14 | Adverse event or side effect | PR | No | 48 | 48 | Alive |
| PT0718 | 5 | Other - resection | SD | No | 51 | 51 | Alive |
| PT0623 | 2 | Other - sepsis, unrelated to protocol treatment | PR | No | 38 | 38 | Dead - brain bleed related to fall |
| PT0655 | 19 | Other - patient and physician decision | uCR | No | 55 | 55 | Alive |
| PT0681 | 17 | Adverse event or side effect | CR | No | 45 | 45 | Alive |
| PT0772 | 22 | Other - patient and physician decision | PR | No | 42 | 42 | Alive |
| PT0770 | 20 | Progression | PR | 20% Increase | 14 | 24 | Dead - melanoma brain metastases |
| PT0777 | 10 | Adverse event or side effect | CR | No | 40 | 40 | Alive |
| PT0796 | 14 | Other - COPD/dyspnea, unrelated to protocol tx | CR | No | 21 | 21 | Dead - cardiac related |
| PT0797 | 25 | Adverse event or side effect | PR | No | 34 | 34 | Alive |
| PT0794 | 15 | Adverse event or side effect | PR | No | 36 | 36 | Alive |
| PT0839 | 34 | Treatment complete | CR | No | 34 | 34 | Alive |
| PT0789 | 34 | Treatment complete | PR | No | 31 | 31 | Alive |
CR, confirmed complete response; PD, progressive disease; PR, confirmed partial response; SD, stable disease; uCR, unconfirmed complete response.
Participant PT0607 withdrew consent after the first cycle of treatment.
Extended Data Table 3 |.
Centralized pathology review of diagnostic biopsies and on-therapy biopsies
| ID | Diagnostic pathologya | On-therapy biopsyb |
|---|---|---|
| PT0522 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0523 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0527 | Not confirmed as desmoplastic melanoma | Some areas of pathological response |
| PT0535 | Confirmed as desmoplastic melanoma | No pathological response |
| PT0561 | Not confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0565 | Biopsy not evaluable | Biopsy not available |
| PT0570 | Confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0582 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0592 | Confirmed as desmoplastic melanoma | No pathological response |
| PT0607 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0623 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0628 | Confirmed as desmoplastic melanoma | No pathological response |
| PT0650 | Confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0655 | Confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0681 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0706 | Biopsy not evaluable | Some areas of pathological response |
| PT0710 | Confirmed as desmoplastic melanoma | No pathological response |
| PT0718 | Confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0723 | Confirmed as desmoplastic melanoma | Some areas of pathological response |
| PT0770 | Not confirmed as desmoplastic melanoma | Biopsy not available |
| PT0772 | Confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0777 | Biopsy not available | Biopsy not available |
| PT0789 | Confirmed as desmoplastic melanoma | Some areas of pathological response |
| PT0794 | Not confirmed as desmoplastic melanoma | Pathological tumor response with no residual viable melanoma cells |
| PT0796 | Biopsy not available | Biopsy not available |
| PT0797 | Confirmed as desmoplastic melanoma | Biopsy not available |
| PT0839 | Confirmed as desmoplastic melanoma | Biopsy not available |
Biopsy of the primary lesion or the S1512 clinical trial baseline biopsy was reviewed centrally to confirm the diagnosis of desmoplastic melanoma. Two pathologists (D.W. and J.A.P.) reviewed the available stained slides (including H&E and immunohistochemistry for either S100 or SOX10 (melanoma cell markers)), resulting in a consensus determination of whether the biopsies were consistent with the diagnosis of desmoplastic melanoma. Consensus annotation was used to describe whether there appeared to be tumor content in the diagnostic biopsies and if the two pathologists agreed that the tumor region of these biopsies was consistent with desmoplastic melanoma pathology.
On-therapy biopsies were collected after one cycle of therapy (between days 28 and 35 after the start of treatment). Slides submitted for central review were analyzed to determine whether there was tumor content and if the two pathologists agreed that the tumor region was consistent with desmoplastic melanoma pathology (consistent with the centralized pathology review of the diagnostic biopsies). In addition, slides were reviewed for a consensus on whether there were features of pathological response by the tumor, including the presence of treatment effect (fibrosis or melanosis) and whether there were viable tumor cells. Biopsy not available: the biopsy sample was not provided for central review. Biopsy not evaluable: the provided biopsy sample did not contain tumor cells or areas that were consistent with a regressed tumor. No pathological response: indicates that there were only viable tumor cells and an absence of a treatment effect by pathological review. Some areas of pathological response: the biopsy had areas consistent with a pathological response, but there were other areas with viable tumor cells remaining. Entirely pathological tumor response (no viable tumor): the biopsy had no viable tumor cells, and the entire resection contained treatment effect with evidence of pathological response.
Extended Data Table 4 |.
Summary data from the centralized pathology review of diagnostic biopsies and on-therapy biopsies
| Timing | Assessment | Number | Percentage of the evaluable samples |
|---|---|---|---|
| Diagnostic biopsies | Biopsy not available | 2 | - |
| Biopsy not evaluable | 2 | - | |
| Not confirmed to be desmoplastic melanoma | 4 | 17% | |
| Confirmed to be desmoplastic melanoma | 19 | 83% | |
| On-therapy biopsies | Biopsy not available | 12 | - |
| Biopsy not evaluable | 0 | - | |
| No pathological response | 4 | 27% | |
| Some areas of pathological response | 4 | 27% | |
| Pathological tumor response with no residual viable melanoma cells | 7 | 46% |
Biopsy not available: the biopsy sample was not provided for central review. Biopsy not evaluable: the provided biopsy sample did not contain tumor cells or areas that were consistent with a regressed tumor. No pathological response: indicates that there were only viable tumor cells and an absence of a treatment effect by pathological review. Some areas of pathological response: the biopsy had areas consistent with a pathological response, but there were other areas with viable tumor cells remaining. Entirely pathological tumor response (no viable tumor): the biopsy had no viable tumor cells, and the entire resection contained treatment effect with evidence of pathological response.
Extended Data Table 5 |.
Treatment-related toxicities observed in ≥5% of patients who received protocol therapy
| Adverse Event | Pembrolizumab (n=27) |
|
|---|---|---|
| Any grade | Grade 3–5 | |
|
| ||
| Fatigue | 15 (56%) | 0 (0%) |
| Diarrhea | 9 (33%) | 2 (7%) |
| Rash maculo-papular | 8 (30%) | 2 (7%) |
| Pruritus | 6 (22%) | 0 (0%) |
| Anemia | 5 (19%) | 0 (0%) |
| Arthralgia | 5 (19%) | 0 (0%) |
| Lymphocyte count decreased | 5 (19%) | 0 (0%) |
| AST increased | 4 (15%) | 0 (0%) |
| Generalized muscle weakness | 4 (15%) | 0 (0%) |
| Headache | 4 (15%) | 0 (0%) |
| Hypothyroidism | 4 (15%) | 0 (0%) |
| Myalgia | 4 (15%) | 0 (0%) |
| Nausea | 4 (15%) | 0 (0%) |
| ALT increased | 3 (11%) | 0 (0%) |
| Blood/lymph disorder - other | 3 (11%) | 0 (0%) |
| Colitis | 3 (11%) | 2 (7%) |
| Constipation | 3 (11%) | 0 (0%) |
| Dyspnea | 3 (11%) | 0 (0%) |
| Hypoalbuminemia | 3 (11%) | 0 (0%) |
| Skin/subcutaneous tissue disorder - other | 3 (11%) | 1 (4%) |
| Adrenal insufficiency | 2 (7%) | 1 (4%) |
| Anorexia | 2 (7%) | 0 (0%) |
| Back pain | 2 (7%) | 1 (4%) |
| Cough | 2 (7%) | 0 (0%) |
| Dizziness | 2 (7%) | 0 (0%) |
| Dry skin | 2 (7%) | 0 (0%) |
| Eye disorder – other | 2 (7%) | 0 (0%) |
| Hyperglycemia | 2 (7%) | 0 (0%) |
| Hyperkalemia | 2 (7%) | 0 (0%) |
| Hypokalemia | 2 (7%) | 1 (4%) |
| Hyponatremia | 2 (7%) | 1 (4%) |
| Pain | 2 (7%) | 0 (0%) |
| Pain in extremity | 2 (7%) | 0 (0%) |
| Pancreatitis | 2 (7%) | 2 (7%) |
| Platelet count decreased | 2 (7%) | 0 (0%) |
All adverse events reported were assessed as possibly, probably or definitely related to the study treatment. Adverse event severity was scored using the National Cancer Institute CTCAE version 4.0. No grade 5 (led to death) adverse events were reported. ALT, alanine transaminase; AST, aspartate aminotransferase.
Extended Data Table 6 |.
Toxicities resulting in serious adverse events (SAEs), determined by the treating physician to be related to the treatment, reported among patients who received protocol therapy
| Adverse Event | Any Grade | Grade 1 | Grade 2 | Grade 3 | Grade 4 | Grade 5 |
|---|---|---|---|---|---|---|
| Adrenal insufficiency | 2 (7%) | 0 | 1 | 1 | 0 | 0 |
| Pancreatitis | 2 (7%) | 0 | 0 | 2 | 0 | 0 |
| Creatinine phosphokinase increased | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Cardiac troponin I increased | 1 (4%) | 1 | 0 | 0 | 0 | 0 |
| Colitis | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Diarrhea | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Fever | 1 (4%) | 0 | 1 | 0 | 0 | 0 |
| Generalized muscle weakness | 1 (4%) | 0 | 1 | 0 | 0 | 0 |
| Immune-mediated duodenitis | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Lung infection | 1 (4%) | 0 | 0 | 0 | 1 | 0 |
| MS/connective tissue disorder | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Myositis | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
| Upper GI hemorrhage | 1 (4%) | 0 | 0 | 1 | 0 | 0 |
Adverse event severity was scored using National Cancer Institute CTCAE version 4.0.
Supplementary Material
Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41591-025-03875-5.
Acknowledgements
The authors would like to acknowledge the participation of patients and their caregivers, the support of patient advocates S. Guild and V. Guild (deceased) and the support of Merck Sharp & Dohme in providing investigational agents for the study. Funding was from SWOG National Institutes of Health/National Cancer Institute (NIH/NCI) grants LS1616_R01LDRGAPP01, U10CA180888, U10CA180819, U10CA180821 and U10CA18068 to S.L.B., S.P.P., J. Moon and M.C.W.A.R. is funded by NIH/NCI grants P01CA244118 and R35CA197633, Agilent Thought Leader Award, the Parker Institute for Cancer Immunotherapy and the Ressler Family Fund and by support from Mary Tanner and Maurizio Grimaldi, Ken and Donna Schultz, Todd and Donna Jones and Mary Jean and Robert (Bob) Rumer. K.M.C. was supported by the Cancer Research Institute Postdoctoral Fellowship Program, the V Foundation Gil Nickel Melanoma Research Fellowship, the Parker Institute for Cancer Immunotherapy and the V Foundation Bridge Fellowship. N.N.A.-D. was funded by the Alan Ghitis Fellowship Award for Melanoma Research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Footnotes
Competing interests
K.L.K.: institutional research support from Bristol Myers Squibb and trial support from GlaxoSmithKline, Immunocore, Varian Medical Systems and Merck. S.L.B. reports no competing interests. Z.E.: advisory board for Regeneron, Pfizer, Replimune, Incyte, Natera and SunPharma and research funding from Pfizer and Boehringer Ingelheim. S.H.-L.: scientific advisor/consultant for Amgen, Ascendis, Astellas, Bristol Myers Squibb, Genmab, Endeavor, Immunocore, Merck, Nektar, Neon Therapeutics, Novartis, Regeneron, Replimune, Vaccinex and Xencor and contracted research through affiliated institutions from Astellas, Aulos Bio, BioAtla, Bristol Myers Squibb, Boehringer Ingelheim, Checkmate, Dragonfly, Erasca, F Star, Genentech, Immunocore, Iovance, Kite Pharma, Lyell, Merck, Nektar, Neon Therapeutics, OncoC4, Pfizer, Plexxikon, Vaccinex, Vedanta and Xencor. K.M.C. reports being a shareholder in Geneoscopy LLC and Georgiamune and has received consulting fees from Geneoscopy LLC, PACT Pharma, Tango Therapeutics, Flagship Labs 81 LLC, the Rare Cancer Research Foundation, the Jaime Leandro Foundation, Noetik and Georgiamune. W.E.C. reports no competing interests. D.A.W.: clinical trial support from Orlucent, Inc. and Blueprint Medicines. J.A.P. reports no competing interests. J.A.S. reports no competing interests. G.K.I.: institutional research grants/contracts from Pfizer, Regneron, Replmune, Bicara, Merck, Georgiamune, Obsidian, Immunocore, Iovance and Xencor; advisory boards for Pfizer, Regeneron, Replimune and Obsidian; and consultant for Pfizer. A.I.: research funding from Merck. J.H.: institutional research grants/contracts from Merck, Bristol Myers Squibb, Iovance, Lyell, Natera, Skyline and Philogen. A.S.B.: advisory board for Deciphera and research funding from Merck. N.I.K.: advisory board for Regeneron, Merck, Replimune, Immunocore, Iovance Biotherapeutics, Novartis, IO Biotech, MyCareGorithm and HUYABIO International; travel support from Castle Biosciences and Regeneron; data safety monitoring board for Incyte and AstraZeneca; scientific advisory board for T-Knife Therapeutics; study steering committee for Bristol Myers Squibb, Nektar, Regeneron and Replimune; common stock in Bellicum Pharmaceuticals and Amarin; and research funding (to institution) from Bristol Myers Squibb, Merck, Regeneron, Replimune, GlaxoSmithKline, Celgene, Novartis, IDEAYA Biosciences, Modulation Therapeutics and HUYABIO International. J. Markowitz: research support from Morphogenesis, Inc. (now TuHURA Biosciences, Inc.) and Merck. M.M.: advisory boards for Merck and Regeneron. G.N. reports no competing interests. S.K.: speakersʼ bureau for Bristol Myers Squibb. G.C.D.: advisory board for Iovance Biotherapeutics. U.S.: consultancy for Astellas, AstraZeneca, Adaptimmune, Exelixis, Gilead, Imvax, Janssen, Pfizer, Seattle Genetics and Sanofi and research funding (to institution) from Janssen, Exelixis and Astellas/Seattle Genetics. T.R. reports no competing interests. B.N.M. reports no competing interests. E.M. reports no competing interests. I.B.-C. reports no competing interests. C.R.G. reports no competing interests. I.P.G. reports no competing interests. A.V.-C. reports no competing interests. J.M.C. reports no competing interests. N.N.A.-D. reports no competing interests. S.P.P.: honoraria for advisory boards, steering committees, data safety monitoring boards or consulting from Bristol Myers Squibb, Cardinal Health, Castle Biosciences, Ideaya, Immatics, IO Biotech, Merck Sharp & Dohme, Novartis, Obsidian, OncoSec, Pfizer, Replimune, Scancell and TriSalus Life Sciences. E.S.: consulting for DE Shaw Research and advisory board for Mallinckrodt Pharmaceuticals. J. Moon reports no competing interests. M.C.W. reports no competing interests. A.R. has received honoraria for consulting for Amgen, Bristol Myers Squibb, Merck, Novartis and Roche-Genentech; is or has been a member of the scientific advisory board and holds stock in Appia, Apricity, Arcus, Compugen, CytomX, ImaginAb, ImmPact, Inspirna, Kite-Gilead, Larkspur, Lyell, Lutris, MapKure, Merus, Synthekine and Tango; and has received research funding from Agilent Technologies and from Bristol Myers Squibb through Stand Up to Cancer (SU2C) and patent royalties from Arsenal Bio.
Extended data is available for this paper at https://doi.org/10.1038/s41591-025-03875-5.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Redacted versions of the S1512 study protocol and statistical analysis plan are available at ClinicalTrials.gov (NCT02775851). Patient-level data, including a data dictionary, will be available within 6 months of publication in accordance with SWOG and National Clinical Trials Network policy and may be requested following the SWOG Cancer Research Network data request procedures (https://www.swog.org/sites/default/files/docs/2019-12/Policy43_0.pdf). Whole-exome sequencing data are available via the database of Genotypes and Phenotypes (dbGaP) at accession phs004123.v1.p1.
References
- 1.Carlino MS, Larkin J & Long GV Immune checkpoint inhibitors in melanoma. Lancet 398, 1002–1014 (2021). [DOI] [PubMed] [Google Scholar]
- 2.Wolchok JD et al. Nivolumab plus ipilimumab in advanced melanoma. N. Engl. J. Med. 369, 122–133 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Tawbi HA et al. Relatlimab and nivolumab versus nivolumab in untreated advanced melanoma. N. Engl. J. Med. 386, 24–34 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hughes TM et al. Desmoplastic melanoma: a review of its pathology and clinical behaviour, and of management recommendations in published guidelines. J. Eur. Acad. Dermatol. Venereol. 35, 1290–1298 (2021). [DOI] [PubMed] [Google Scholar]
- 5.Shain AH et al. Exome sequencing of desmoplastic melanoma identifies recurrent NFKBIE promoter mutations and diverse activating mutations in the MAPK pathway. Nat. Genet. 47, 1194–1199 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Eroglu Z et al. High response rate to PD-1 blockade in desmoplastic melanomas. Nature 553, 347–350 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Taube JM et al. Colocalization of inflammatory response with B7-H1 expression in human melanocytic lesions supports an adaptive resistance mechanism of immune escape. Sci. Transl. Med. 4, 127ra137 (2012). [Google Scholar]
- 8.Tumeh PC et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 515, 568–571 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Rizvi NA et al. Mutational landscape determines sensitivity to PD-1 blockade in non–small cell lung cancer. Science 348, 124–128 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.McGranahan N et al. Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science 351, 1463–1469 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ribas A & Wolchok JD Cancer immunotherapy using checkpoint blockade. Science 359, 1350–1355 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Campbell KM et al. Prior anti-CTLA-4 therapy impacts molecular characteristics associated with anti-PD-1 response in advanced melanoma. Cancer Cell 41, 791–806 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ansell SM et al. PD-1 blockade with nivolumab in relapsed or refractory Hodgkinʼs lymphoma. N. Engl. J. Med. 372, 311–319 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Le DT et al. PD-1 blockade in tumors with mismatch-repair deficiency. N. Engl. J. Med. 372, 2509–2520 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Le DT et al. Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade. Science 357, 409–413 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nghiem PT et al. PD-1 blockade with pembrolizumab in advanced Merkel-cell carcinoma. N. Engl. J. Med. 374, 2542–2552 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cercek A et al. PD-1 blockade in mismatch repair-deficient, locally advanced rectal cancer. N. Engl. J. Med. 386, 2363–2376 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Goh G et al. Mutational landscape of MCPyV-positive and MCPyV-negative Merkel cell carcinomas with implications for immunotherapy. Oncotarget 7, 3403–3415 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Murali R et al. Prognostic factors in cutaneous desmoplastic melanoma: a study of 252 patients. Cancer 116, 4130–4138 (2010). [DOI] [PubMed] [Google Scholar]
- 20.Han D et al. Clinicopathologic predictors of survival in patients with desmoplastic melanoma. PLoS ONE 10, e0119716 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Brahmer JR et al. Safety profile of pembrolizumab monotherapy based on an aggregate safety evaluation of 8937 patients. Eur. J. Cancer 199, 113530 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Posther KE et al. Histopathologic characteristics, recurrence patterns, and survival of 129 patients with desmoplastic melanoma. Ann. Surg. Oncol. 13, 728–739 (2006). [DOI] [PubMed] [Google Scholar]
- 23.Hughes TM et al. Desmoplastic melanoma: a review of its pathology and clinical behaviour, and of management recommendations in published guidelines. J. Eur. Acad. Dermatol. Venereol. 35, 1290–1298 (2001). [Google Scholar]
- 24.Clopper C & Pearson E The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika 26, 404–413 (1934). [Google Scholar]
- 25.Pintilie M Competing Risks: A Practical Perspective (John Wiley & Sons, 2006). [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Redacted versions of the S1512 study protocol and statistical analysis plan are available at ClinicalTrials.gov (NCT02775851). Patient-level data, including a data dictionary, will be available within 6 months of publication in accordance with SWOG and National Clinical Trials Network policy and may be requested following the SWOG Cancer Research Network data request procedures (https://www.swog.org/sites/default/files/docs/2019-12/Policy43_0.pdf). Whole-exome sequencing data are available via the database of Genotypes and Phenotypes (dbGaP) at accession phs004123.v1.p1.