Efficacy of PARP inhibitor therapy after targeted BRAF/MEK failure in advanced melanoma

Abstract

Modern advancements in targeted therapy and immunotherapy have significantly improved survival outcomes for advanced melanoma; however, there remains a need for novel approaches to overcome disease progression and treatment resistance. In recent years, PARPi therapy has shown great promise both as a single regimen and in combination with other therapeutics in melanoma. Here, we describe three unique cases of advanced BRAF V600 mutated melanoma that progressed on targeted BRAF/MEK agents that subsequently exhibited partial to near-complete responses to combinatory PARPi and BRAF/MEK inhibitors. This highlights both a potential synergy underlying this combinatory approach and its efficacy as a treatment option for patients with advanced melanoma refractory to targeted and/or immunotherapies. Prospective clinical trials are needed to explore this synergic effect in larger melanoma cohorts to investigate this combination for treating refractory advanced melanoma.

Introduction

Melanoma, the deadliest type of skin cancer, accounts for nearly 2% of all global cancer diagnoses and remains the fifth most common cancer in the US¹. Alarmingly, the prevalence of the disease has increased drastically in recent years, with over 100,000 new cases and 8,000 deaths expected in 2024². However, melanoma mortality, especially from advanced/metastatic disease, has dropped significantly due to modern advancements in immunotherapy and targeted therapies (e.g., BRAF, MEK, MAPK)^1–4. Activating mutations in BRAF, which are present in about 50% of metastatic melanoma, are recognized as predictive biomarkers for targeted therapies in patients with advanced melanoma. Furthermore, nearly 50% of patients with advanced melanoma treated with immunotherapies, such as combination ipilimumab and nivolumab, remain alive at 7.5 years, a vast improvement over the prior era of cytotoxic chemotherapies and cytokine treatment⁵. In addition, the NEMO trial demonstrated that binimetinib, a MEK inhibitor, contributed to a significantly longer progression-free survival (2.8 months) compared to dacarbazine (1.5 months) in patients with NRAS-mutant advanced melanoma⁶. However, resistance and intolerance to both immunotherapy and targeted therapies are common and remain a challenge in advanced melanoma treatment^7–9. Thus, novel therapeutics or combinatory approaches are required to overcome resistance and disease progression, especially in cases of advanced disease.

Poly (ADP-Ribose) Polymerase inhibitors (PARPi) have emerged as a promising treatment in cancers with alterations in homologous recombination repair (HRR) genes, a type of DNA damage repair (also termed HR-DDR). PARPi treatments work by creating double-stranded DNA (dsDNA) breaks that, under normal physiology, are repaired via homologous recombination (HR) mechanisms. However, homologous recombination deficient (HRD) cells cannot repair dsDNA breaks and thus undergo cell death via synthetic lethality. HRD can occur due to the presence of HR-DDR pathway alterations, such as gene mutations in BRCA1, BRCA2, ATM, CHEK2, and PALB2^10–12. HR-DDR mutations in melanoma are common, with frequencies ranging from 18.1% to upwards of 41%^13,14. In general, patients with advanced melanoma without HR-DDR mutations respond poorly to PARPi^15–18, however, there are reports of favorable responses to PARPi in patients harboring HR-DDR mutations^17–19. Furthermore, the addition of PARPi to immune checkpoint inhibitors (ICI), such as anti-PD1 therapy, has been proposed due to the immune priming effects of PARPi therapy, such as increasing PD-L1 expression and genomic instability, which further promote immune response to ICI and antitumor activity^20,21. Clinically, this synergism was seen in a study that observed combinatory ICI and PARPi therapy to demonstrate a consistent objective response rate in a cohort of patients with ovarian cancer²². In addition, the synergism between PARPi and BRAF/MEK inhibition has been explored preclinically in melanoma showing promise in using this combinatory approach^23,24. These findings warrant further research on the efficacy and safety of PARPi as standalone therapy or in combination with other therapeutic options in treating patients with melanoma.

Here, we report three unique cases of advanced melanoma refractory to BRAF/MEK inhibition that subsequently demonstrated favorable clinical responses to PARPi therapy. We also assessed treatment response through longitudinal, tumor-informed circulating tumor DNA (ctDNA) monitoring when possible. These results help contextualize the efficacy of combinatory PARPi and targeted therapy within a framework of overcoming treatment resistance in patients with advanced melanoma.

Results

We identified three patients at Siteman Cancer Center of Washington University School of Medicine who presented for management of metastatic melanoma. All three patients harbored an activating mutation in the MAP kinase pathway and most (two of the patients) harbored an alteration in HR pathways. All patients initially progressed on immunotherapy as well as BRAF/MEK inhibitor (BRAF/MEKi) therapy, after which a PARPi was added with significant clinical responses (one complete response [CR] and two partial responses [PR] based on Recist 1.1 criteria²⁵). Treatments were well-tolerated. Patient demographics and clinical data are reported in Table 1.

Table 1.

Patient demographics and clinical data

	Patient 1	Patient 2	Patient 3
Demographics	60-year-old, Male, Caucasian (non-Hispanic)	70-year-old, Male, Caucasian (non-Hispanic)	55-year-old, Female, Caucasian (ethnicity not specified)
PMHx	None	Barrett’s esophagus, hypertension, hyperlipidemia, adrenal adenoma (unchanged)	Migraines, hypertension
Family History	No melanoma or non-melanoma skin cancers	No melanoma or non-melanoma skin cancers	No melanoma or non-melanoma skin cancers
Melanoma Location and Pathology	Primary: right shin Metastasis: LN (right iliac, retroperitoneal) Subtype: nodular	Primary: mid-upper back Metastasis: brain, lung, heart, adrenal glands, superficial soft tissue/muscle and pelvic Subtype: NOS	Primary: posterior right calf Metastasis: LN (right femoral) and lung Subtype: nodular
Stage at diagnosis	Stage IIIC (pT3, pN3b, M0)	Stage IV	Stage IIB (pT3b, N0)
Therapies before BRAF/MEKi therapy and response	1. WLE and SLNB 2. Adjuvant pembrolizumab 200 mg q3 weeks then recurrence after 33.2 months 3. Ipilimumab 3 mg/kg q3 weeks /nivolumab 1 mg/kg q2 weeks for 0.7mths (discontinued due to grade 3 ICI-associated CIPD) 4. Resection and retroperitoneal LN dissection	1. Ipilimumab 3 mg/kg q3 weeks + nivolumab 1 mg/kg q4 weeks (discontinued after 1 month due to grade 3 ICI-associated pneumonitis)	1. WLE of right leg and SLNB 2. Resection of right femoral LN due to recurrence (Stage IIIC) 3. High dose adjuvant ipilimumab 10 mg/kg q3 weeks (PD after 3 months) 4. Pembrolizumab 200 mg q3 weeks for 19mths (no response) 5. Atezolizumab 840 mg q2 weeks + oral cobimetinib 40 mg PO QD on clinical trial (PR) 6. Atezolizumab 1200 mg every 3 weeks off trial for 4.5mths
BRAF/MEKi therapy	Regimen: binimetinib 45 mg PO BID Duration: 9.1 months Response: PD AE: acneiform rash	Regimen: encorafenib 450 mg PO QD + binimetinib 45 mg PO BID Duration: 9.5 months Response: MR AE: hypothyroidism	Regimen: encorafenib 450 mg PO QD + binimetinib 45 mg PO BID Duration: 2.3 months Response: MR AE: drug-induced diabetes mellitus
Interval therapy	None	1. Gamma knife to the right parietal lesion 2. LITT to the right temporal and occipital lobe lesions	1. Nivolumab 1 mg/kg q4 weeks + encorafenib/binimetinib (MR after 3.7 months) 2. interval T-VEC + nivolumab (PD after 4.6 months) 3. Restarted nivolumab + encorafenib/binimetinib (PD as a new left lower lobe pulmonary nodule after 18 months)
PARPi therapy	Regimen: olaparib 300 mg PO BID + binimetinib, radiation therapy to abdominal lymph nodes Time to response: 6 months Duration: 15 months and ongoing Response: CR Adverse effect: none	Regimen: olaparib 300 mg PO BID + encorafenib/binimetinib with intermittent brain radiation therapy and LITT Time to response: 2.6 months Duration: 12.6 months and ongoing Response: PR Adverse effect: none	Regimen: olaparib 300 mg PO BID + encorafenib/binimetinib + nivolumab Time to response: 3.5 months Duration: 7 months and ongoing Response: PR Adverse effect: nausea and fatigue (Olaparib reduced to every other day)

PMHx past medical history, BRAF/MEKi BRAF/MEK inhibitor, PARPi PARP inhibitor, PR partial response, CR complete response, NOS not otherwise specified, PD progressive disease, MR mixed response, WLE wide local excision, SLNB sentinel lymph node biopsy, LN lymph node, ICI immune checkpoint inhibitor, CIPD chronic inflammatory demyelinating polyradiculoneuropathy, LITT laser interstitial thermal therapy, T-VEC talimogene-laherparepvec, AE adverse effects.

Patient 1

A 60-year-old Caucasian male presented for management of a primary nodular melanoma of the right shin with lymph node (right iliac and retroperitoneal) metastases. He noticed a pigmented lesion on his right shin which ulcerated and underwent wide local excision (WLE) and sentinel lymph node biopsy (SLNB) which revealed Stage IIIC (pT3 pN3b M0) disease. Molecular profiling detected mutations in NRAS, TERT promoter, EZH2, ARID2, and CHD2 with a low genomic loss of heterozygosity (LOH) and tumor mutational burden (TMB) (Table 2). Before MEK inhibitor monotherapy, he received adjuvant pembrolizumab (200 mg q3 weeks), followed by ipilimumab (3 mg/kg q3 weeks) /nivolumab (1 mg/kg q2 weeks) due to metastatic recurrence in the right iliac and retroperitoneal lymph nodes. Unfortunately, after 2 cycles, he developed grade 3 immune checkpoint inhibitor (ICI)-associated neuropathy and proceeded to surgical resection. Due to the NRAS mutation, the patient was subsequently started on binimetinib (45 mg PO BID) with no evidence of radiographic response by 9.1 months and persistent ctDNA-positivity (Fig. 1A). Olaparib(300 mg PO BID), a PARPi, was added to binimetinib along with radiation therapy (35 Gy in 5 fractions) to abdominal lymph nodes, resulting in a partial response (PR) after 6 months and a complete response (CR) after 15 months, which has persisted at last follow up and correlated with sustained ctDNA clearance (Fig. 1A). No serious adverse events to MEKi or PARPi therapy occurred.

Table 2.

Melanoma tumor genetic and molecular profiling

	Patient 1	Patient 2	Patient 3
Mutations	NRAS- pathogenic variant p.Q61K, exon 3; VAF 37% TERT promoter- pathologic variant; VAF 33% EZH2- pathological variant p.Y646N, exon 16; VAF 32% ARID2- pathogenic variant p.W264, exon 8; VAF 17%	BRAF- pathologic variant p.V600K, exon 15; VAF 70% ATM-deletion IDH1-pathologic variant p.R132C, exon 4; VAF 45% MGA-pathologic variant p.R2396, exon 19; VAF 43% MRE11-deletion TERT promoter-pathologic variant, exon 0; VAF 63% ARID2- variant of uncertain significance p.R652K, exon 15; VAF 38% RET- variant of uncertain significance p.5444 F, exon 7; VAF 42% POLE-deletion	BRAF- pathologic variant p.V600K, exon 15; VAF 33% NTRK1- variant of uncertain significance p.G307E, exon 8; VAF 21% TERT promoter-pathogenic variant; VAF 67% B2M-pathogenic variant p.L85fs, exon 2; VAF 39% WRN-pathogenic variant p.Q1010, exon 25; VAF 19%
MSI	Stable	Stable	Stable
MMR status	Intact	Intact	Intact
TMB (mutations per Mb)	Low (6)	High (21)	High (11)
LOH	Low (8%)	Low (5%)	Low (12%)
PD-L1	80%	0%	0%

TMB tumor mutational burden, MSI microsatellite instability, MMR mismatch repair, LOH loss of heterozygosity, PD-L1 Programmed death ligand 1, VAF Variant allele frequency, Mb megabase.

Fig. 1. Timeline correlating treatment and clinical response (using ctDNA and imaging) for patients 1 and 3.

A Patient 1. Top: CT-imaging 3 months after Olaparib initiation demonstrating resolution of a right common iliac node (previously 17 mm in short axis). Bottom: Graph depicting a timeline of administered systemic therapy and ctDNA changes during surveillance. White circle – ctDNA-negative; Black circle – ctDNA-positive; red triangle – progressive disease on imaging; light green triangle – partial response on imaging; dark green triangle – complete response on imaging; light purple rectangle – Binimetinib; light blue rectangle – combinatory Binimetinib and Olaparib; vertical dark blue line – initiation of radiotherapy. B Patient 3. Top: CT-imaging 3 months after Olaparib initiation demonstrating interval decrease in the size of a left lower lobe pulmonary nodule measuring 1.1 cm (previously measured 1.4 cm). Bottom: Graph depicting a timeline of administered systemic therapy and ctDNA changes during surveillance. Black circle – ctDNA-positive; red triangle – progressive disease on imaging; light green triangle – partial response on imaging; dark green triangle – complete response on imaging; light yellow rectangle – combinatory T-VEC and Nivolumab; purple rectangle – combinatory Encorafenib, Binimetinib, and Nivolumab; bright blue rectangle – combinatory Olaparib, Encorafenib, Binimetinib, and Nivolumab.

Patient 2

A 70-year-old Caucasian male presented for management of a mid-upper back primary melanoma with widespread metastases (Table 1). He presented to the emergency department with aphasia and was found to have a left temporal brain lesion with biopsy revealing malignant melanoma (type not otherwise specified). Notably, a mutation in ATM (HR-DDR gene) was present amongst other mutations including BRAF V600K, TERT promoter, and ARID2; LOH was low and TMB was high (Table 2). He started treatment with ipilimumab (3 mg/kg q3 weeks)/nivolumab (1 mg/kg q4 weeks); however, this was discontinued after one cycle due to grade 3 ICI-associated pneumonitis. Encorafenib (450 mg PO QD)/binimetinib (45 mg PO BID) was then initiated with an overall mixed response and treatment-associated hypothyroidism after 9.5 months of therapy. He then underwent treatment of three brain metastases (gamma knife radiosurgery for parietal lesion; laser interstitial thermal therapy (LITT) for right temporal and occipital lesions). He then started olaparib (300 mg PO BID) and radiation therapy (stereotactic radiosurgery [SRS, 1 fraction, 21 Gy] to a new frontal lesion; LITT and fractionated SRS [5 fractions, 30 Gy] to right occipital and parietal lesions). He achieved a partial response after 2.6 months of PARPi therapy with intermittent local therapies to brain metastases, which has persisted at his last follow-up (12.6 months). No other adverse events to BRAF/MEKi or PARPi therapy occurred.

Patient 3

A 55-year-old Caucasian female presented for management of a Stage IIB nodular melanoma of the right calf. She underwent WLE and SLNB but later had recurrence in the right femoral lymph nodes that was treated with lymph node dissection and high-dose adjuvant ipilimumab (10 mg/kg q3 weeks) for BRAF wild-type disease. Unfortunately, she developed a second recurrence (new right groin mass) which was refractory to pembrolizumab (200 mg every 3 weeks) after 19 months of therapy. Subsequently, she was enrolled in a clinical trial with atezolizumab (840 mg q2 weeks)/cobimetinib (40 mg PO QD) followed by off-trial atezolizumab (1200 mg q3 weeks) due to clinical benefit seen despite trial closure. A repeat lymph node biopsy revealed tumor mutations including BRAF V600K, NTRK1, TERT promoter, B2M, and WRN (HR-DDR gene) with a low LOH and high TMB (Table 2). Due to disease progression and a newly acquired BRAF mutation, she was started on encorafenib (450 mg PO QD)/ binimetinib (45 mg PO BID) but had a mixed response (after 2.3 months), thus nivolumab (1 mg/kg q4 weeks) was added to her regimen. She still only showed a mixed response after 3.7 months of therapy, so BRAF/MEKi was held while she received nivolumab with intralesional talimogene-laherparepvec (T-VEC), to which she experienced disease progression (after an additional 4.6 months). She then restarted combination therapy with nivolumab and encorafenib/binimetinib (as it was thought to have contained her disease), however, she again experienced disease progression (new left lower lobe pulmonary nodule) with a concurrent rise in ctDNA levels after 18 months (Fig. 1B). Consequently, olaparib (300 mg PO BID) was added to her regimen, and she achieved a partial response after 3.5 months along with a marked reduction in ctDNA levels (Fig. 1B), which has persisted at last follow-up. Notably, she was not taking her Olaparib as prescribed for two months (due to associated nausea and fatigue), so it was reduced to every other day at the last follow-up visit. No other adverse events to BRAF/MEKi or PARPi therapy were reported.

Discussion

This case series reviewed three patients with advanced melanoma refractory to targeted therapy (BRAF/MEK inhibition) who subsequently responded to the addition of PARPi therapy (2 PR, 1 CR). Treatment was generally well-tolerated, and responses were seen rapidly within 2.6–6 months of initiating PARPi therapy.

PARPi is an emerging class of cancer therapeutics, and patient selection remains crucial in personalizing care and overcoming prior treatment resistance. HR-DDR pathway gene mutations (e.g., BRCA1/2, CHEK2, ATM – see Table 3 for a comprehensive list) and high HRD status (measured as a composite score from a single or multiple genomic instability measures such as LOH²⁶) have been identified as biomarkers for PARPi therapy in patients with melanoma. We have previously described successful clinical responses to PARPi in melanoma harboring HR-DDR gene mutations¹⁹ or high HRD status without HR-DDR gene mutations^27,28. Additionally, we previously presented an advanced acral melanoma patient without traditional HR-DDR mutations achieving a complete response to PARPi following ICI failure. This patient had an EMSY amplification, which has been implicated in being involved in the HR-DDR pathway²⁹. These reports suggest that a multifactorial basis underlies the response to PARPi therapy in melanoma. Interestingly, in the present study, all patients had low HRD status (as determined by genomic LOH), yet all patients still achieved significant clinical responses upon the addition of PARPi therapy, which corresponded to ctDNA clearance or a marked decrease in ctDNA levels when such testing was performed. Importantly, patients 2 and 3 harbored mutations in ATM and WRN, respectively, which are less commonly involved in the HR-DDR pathway and could potentially help explain the partial response to combinatory treatment. Nonetheless, these results provide real-world experience supporting the use of PARPi in patients with HR-DDR pathway gene mutations. Furthermore, these results provide clinical experience supporting a possible synergistic interaction between PARPi and targeted therapy (e.g., BRAF/MEK inhibition) in patients with advanced melanoma. However, more studies are needed to confirm this synergism preclinically and clinically in larger cohorts.

Table 3.

Common mutations affecting homologous recombination pathways

Homologous Recombination DNA Damage Response
Genes29,45	ARID1A, ARID1B, ATM, ATR, ATRX, BAP1, BARD1, BLM, BRCA1, BRCA2, BRIP1, CHEK2, EMSY, FANCA/C/D2/E/F/G/L, MRE11, NBN, PALB2, RAD50, RAD51, RAD51B/C, WRN

The synergism of PARPi and BRAK/MEK inhibitors is not well described in the literature, however, several preclinical studies evaluate this relationship. Frohlich et al. demonstrated that melanoma cells resistant to MAPK inhibitors (MAPKi) were highly susceptible to PARPi treatment in vitro and in vivo via a synergistic effect of synthetic lethality, causing reduced melanoma cell proliferation and migration²³. Interestingly, the authors observed that MAPKi-resistant melanoma cells had exhibited lower basal ATM expression, theorizing that these cells had decreased HRR activity leading to a reduced repair of dsDNA caused by PARPi and suggesting melanoma cell ATM expression as a potential novel biomarker for PARPi therapy response²³. Furthermore, Ferretti et al. observed that PARPi treatment had restored sensitivity to MAPKi in melanoma cells resistant to MAPKi regardless of DDR²⁴. The authors noted that PARPi treatment causes transcriptomic and epigenetic changes that reverse epithelial-mesenchymal transition-like phenotype switching, a leading cause for therapeutic resistance^24,30–32, redirecting melanoma cells to a proliferative and sensitive state to MAPKi. Additionally, Maertens et al. demonstrated that MAPKi suppressed several HR pathway genes in MAPKi-sensitive melanoma cells, ultimately inducing an increased HR-defect signature in treated cells. Importantly, this BRCA-ness phenotype is a biomarker for PARPi, supported by the authors observing increased cytotoxicity in melanoma MAPKi-sensitive cell lines after subsequent PARPi treatment (versus no potent cytotoxic effect when using PARPi monotherapy)³³.

Furthermore, Maertens et al. analyzed transcriptional data between MAPKi-sensitive melanoma cell lines and Histone deacetylase (HDAC) inhibitors and found elevated MGMT gene expression (9-fold). Moreover, they discovered that MGMT-expressing cells were sensitive to this combinatory treatment, whereas cell lines lacking MGMT expression were not responsive to this treatment approach. The authors highlighted that MGMT expression could be utilized as a biomarker for response to combinatory MAPKi and HDAC inhibitors. Additionally, they observed high MGMT-expressing melanomas to possess broad DNA repair gene defects, and ultimately, an HRD gene signature³³. Unfortunately, MGMT expression for the patients in this study is not available. Nonetheless, future studies are warranted to elucidate the role of MGMT as a biomarker in melanoma response to combinatory MAPKi and PARPi.

These preclinical findings, together with our clinical results, strongly suggest a synergistic effect of MAPKi and PARPi as a combinatory treatment option in patients with advanced melanoma, especially those with immunotherapy resistance. In addition, PARPi were found to increase cell-intrinsic immunity by creating cytoplasmic chromatin fragments that further activate the cGAS/STING pathway, increase cytokine production, and activate downstream IFN signaling in DNA damage response-deficient tumor cells³⁴. Moreover, PARPi have shown promise as radiosensitizers in patients with various cancers, including breast cancer and melanoma^35,36, which could also help explain the responses demonstrated by patients 1 and 2.

Interestingly, patients 1 and 2 harbored ARID2 mutations, which is a member of the SWI/SNF family of chromatin remodeling complexes. ARID2 was implicated to be indirectly involved in the DDR pathways^37–39, and deficiencies in the gene have been associated with response to ICI in melanoma⁴⁰. Interestingly, Moreno et al. highlighted in a study how lung cancer cells with ARID2 deficiencies were associated with sensitivity to PARPi treatment⁴¹. These results highlight the potential utility of ARID2 as a biomarker for PARPi response, however, more investigation is needed to confirm these results in larger melanoma cohorts.

In conclusion, the results of this study support a growing body of evidence on the efficacy and safety of using PARPi in treating advanced melanoma that is refractory to targeted BRAF/MEK inhibition and/or ICI. Our study is limited by a small sample size and its retrospective nature. Larger, randomized controlled trials are required to further elucidate synergistic mechanisms and establish the potential benefit of PARPi in patients with advanced melanoma refractory to immunotherapy and targeted therapies.

Methods

Materials

Formalin-fixed paraffin-embedded (FFPE) samples were prepared from patient biopsies and submitted to a commercial, CLIA-certified laboratory for molecular profiling (Caris Life Sciences, Phoenix, AZ). Samples were analyzed by next-generation sequencing (NGS) of DNA (whole exome sequencing, WES) and RNA (whole transcriptome sequencing, WTS), along with immunohistochemistry (IHC).

DNA next generation sequencing (NGS)

In preparation of the samples for molecular testing, tumor enrichment was done by harvesting targeted tissues using manual microdissection techniques. Genomic DNA was extracted from FFPE tissue samples and subjected to NGS using the NovaSeq 6000 Platforms (Illumina, Inc. San Diego, CA). A custom SureSelect XT assay (Agilent Technologies, Santa Clara, CA) was utilized to enrich exonic regions 592 whole-gene targets. For tumor sample sequenced on the Novaseq 6000 platform, more than 700 clinically relevant genes were assessed. All variants were detected with >99% confidence based on allele frequency and amplicon coverage, with an average sequencing depth of coverage of >500 and an analytic sensitivity threshold established of 5% for variant calling. Certified molecular geneticists examined the identified genomic variants and categorized them in alignment with the standards set by the American College of Medical Genetics and Genomics (ACMG).

TMB was measured by counting all non-synonymous missense, nonsense, in-frame insertion/deletion, and frameshift mutations found per tumor that had not been previously described as germline alterations in dbSNP151, Genome Aggregation Database (gnomAD) databases, or benign variants identified by Caris’s geneticists. High TMB (TMB-H) was defined by a cut-off of ≥10 mutation/megabase (mut/MB) based on the KEYNOTE-158 pembrolizumab trial, where it was shown that patients with ≥10 mut/MB had increased response rates compared to those with <10 mut/MB⁴². To calculate the genomic loss of heterozygosity (LOH), LOH in approximately 250k single nucleotide polymorphisms (SNPs) within segmented autosomal chromosomes was calculated. LOH was based on the percentage of all 552 segments with observed LOH (High ≥16%, Low <16%; if fewer than 3000 SNPs were read, the test was reported as indeterminate).

Whole transcriptomic sequencing

Formalin-fixed paraffin-embedded (FFPE) tissue sections mounted on glass slides underwent staining with nuclear fast red (NFR). Regions that contained a minimum of 10% tumor content were delineated for manual microdissection and subsequent mRNA extraction. Whole transcriptome sequencing (WTS) was executed using the Illumina NovaSeq platform (Illumina, Inc., San Diego, CA) along with the Agilent SureSelect Human All Exon V7 bait panel (Agilent Technologies, Santa Clara, CA), and the resulting data reported transcripts per million (TPM).

Immunohistochemistry (IHC)

IHC was conducted on complete sections of formalin-fixed paraffin-embedded (FFPE) tissues mounted on glass slides. The slides underwent staining employing automated staining methods as directed by the manufacturer. These procedures were meticulously optimized and confirmed to meet the standards outlined by CLIA/CAO and ISO. PD-L1 expression was determined using primary antibody SP142 (Spring Biosciences, Pleasanton, CA, USA), with a positive threshold of ≥2 + stain intensity and ≥ 5% percentage of cells stained.

Deficient mismatch repair/microsatellite instability-high (dMMR/MSI-H)

dMMR/MSI-H was determined by a combination of immunohistochemistry (IHC) using antibodies for MLH1 (M1 antibody), MSH2 (G2191129 antibody), MSH6 (44 antibody), and PMS2 (EPR3947 antibody) from Ventana Medical Systems (Tucson, AZ), and next-generation sequencing (NGS). The outcomes from these platforms are mostly in agreement, as previously described⁴³. In instances where conflicting results emerged, the order of priority for determining the MSI/MMR status of the tumor was IHC, followed by NGS.

Personalized, tumor-informed ctDNA testing

Longitudinal ctDNA testing was performed in two of the three patients at the discretion of the treating provider. A personalized, tumor-informed, 16-plex PCR assay (SignateraTM, Natera, Inc.) was used for the detection and quantification of ctDNA, as previously described⁴⁴. Briefly, whole-exome sequencing (WES) was performed on FFPE tumor tissue and matched normal blood samples. A set of up to 16 patient-specific somatic single nucleotide variants (SNVs) from WES results were selected for multiplex PCR (mPCR). The mPCR primers targeting the personalized SNVs were used to track ctDNA in the corresponding patients’ plasma samples. Plasma samples with ≥ 2 SNVs detected above a predefined confidence threshold were defined as ctDNA-positive. ctDNA concentration was reported in mean tumor molecules (MTM)/mL of plasma.

Patient consent

The authors have obtained written patient consent from all patients to publish the details in this study. This study was approved by the Washington University IRB.

Author contributions

J.P., G.N., and R.M. performed literature review, extracted/analyzed patient data, and wrote/revised the manuscript. T.A., A.E., F.A., A.V., S.P., O.B. provided expertise in editing and revising the manuscript. G.A. conceived the project idea and approved the final manuscript.

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.