Leveraging homologous recombination deficiency for sarcoma

Defects in homologous recombination repair (HRR) are common in several cancer types but remain poorly understood in sarcoma, a rare and heterogeneous cancer of mesenchymal origin [1, 8]. Determining HRR deficiency is of high clinical relevance as it is associated with susceptibility to poly (ADP-ribose) polymerase (PARP) inhibition [11, 13]. Standard-of-care treatment for advanced-disease sarcoma mostly relies on chemotherapy. Unfortunately, patients often show chemoresistance, and metastatic disease is associated with poor survival [1, 12]. Therefore, identifying underlying disease mechanisms for the distinct sarcoma entities is likely to unlock new therapeutic avenues.

Homologous recombination deficiency (HRD) is frequently observed in ovarian and breast cancer, followed by prostate and pancreatic cancer. Accurate detection of HRD is of clinical relevance as it is indicative of sensitivity to targeted therapy with PARP inhibitors (PARPi) and DNA-damaging agents [13]. The most clinically widespread genetic biomarker of HRD is germline or somatic BRCA1/2 mutation status, hereafter termed BRCAness [8]. However, the prevalence of HRD extends beyond BRCA1/2 inactivation [11]; therefore, the term HRDness will be used throughout this article to encompass tumor-agnostic HRR pathway deficiencies.

To identify possible genomic traits of HRDness in sarcoma, we analyzed signatures of genomic instability in the TCGA cohort of 247 soft tissue sarcoma (STS) cases and the TARGET cohort of 69 osteosarcoma cases [1]. For each individual case, we computed loss-of-heterozygosity (LOH), large-scale transitions (LST), and telomeric allelic imbalances (TAI) as well as the HRD score as the unweighted sum of all three values [9]. The HRD score is a clinically validated biomarker that predicts PARPi and platinum sensitivity in high-grade serous ovarian and triple-negative breast carcinoma (HGSOC and TNBC, respectively) [4, 15], which were used as controls in our analysis. Sarcoma entities known to present high levels of genomic instability, namely myxofibrosarcoma, undifferentiated pleomorphic sarcoma, and osteosarcoma, also exhibited the highest HRD scores, followed by uterine leiomyosarcoma, malignant peripheral nerve sheath tumor, dedifferentiated liposarcoma, and non-uterine leiomyosarcoma. Both synovial sarcoma and desmoid tumors, associated with chromosomal translocations or point mutations in specific genes, presented low HRD scores (Fig. 1a).

Bild vergrößern

A thorough analysis of the molecular landscape of STS identified numerous alterations in HRR pathway genes: 22% of STS patients carried mutations in BRCA2, and 37% and 20% in the Fanconi anemia genes FANCB and FANCA, respectively. The TP53 regulator MDM2, the tumor suppressor PTEN, cell cycle checkpoint regulators RAD1 and CHEK1, and the DNA damage and replication proteins ATM, RPA1, and H2AFX completed the top 10 altered gene list in the TCGA-SARC cohort. As the total number of HRR alterations, referred to as HRR-CIN, followed a bimodal distribution, we applied a finite mixture model that separated the cohort into two subpopulations. The HRD score for each datapoint was associated with the respective HRR-CIN group in a receiver operator characteristic (ROC) curve, and an optimal cut-off value for stratifying patients based on HRD score was determined at 32 (Fig. 1b). Of note, an HRD score above 42 has been clinically validated to predict the response to platinum-containing neoadjuvant chemotherapy in patients with TNBC and recurrent ovarian cancers treated with niraparib [4, 15]. Currently, no clinically validated diagnostic test exists for drug response prediction in soft tissue or bone sarcoma. Besides elevated levels of genomic alterations in STS with high HRD scores (HRD^high), a high degree of chromosomal instability (CIN), majorly influenced by high levels of chromosomal gains, as well as molecular signatures associated with deficiency in HRR were identified [14]. Altogether, we identified distinct sarcoma entities with multiple traits of HRDness, thus building the rationale for PARPi therapy.

To explore the transcriptomic landscape of sarcoma with HRDness, we investigated gene expression profiles and signaling pathway enrichment in HRD^high compared with HRD^low STS patients from the TCGA cohort. Sarcomas exhibiting HRDness showed a general enrichment of DNA damage repair pathways, including HRR and mismatch and nucleotide excision repair, as well as significant upregulation of 10 key HRR genes: BRCA1, BRCA2, BLM, EME1, FANCB, FANCD2, FANCI, RAD51, RAD54L, and XRCC2, which we named the SARC-HRD signature (Fig. 1c). Interestingly, these 10 HRR genes were also upregulated in five HRD^high compared with five HRD^low patient-derived ex vivo sarcoma cell models, concomitantly with increased genomic instability traits, such as high CIN and an elevated number of genomic alterations in HRR genes. Our combined results show that STSs harbor numerous genomic traits of HRDness that are shared with transcriptional upregulation of a distinct gene expression signature.

PARP1 and PARP2 are core players in base-excision-mediated DNA repair (BER). PARPi block the BER pathway and HRR-deficient cells must rely on the error-prone, non-homologous end-joining pathway for double-strand DNA repair. Cells accumulate genetic damage that ultimately induces cell death [11]. In vitro sensitivity to DNA double-strand break-inducing drugs, such as platinum salts, is also a feature of HRD cells. To investigate whether sarcoma with HRDness traits respond to PARPi and platinum chemotherapy, we explored the sensitivity of five HRD^high and five HRD^low patient-derived ex vivo sarcoma cell models to two PARPi (olaparib and niraparib), a platinum-based antineoplastic drug (oxaliplatin), and standard chemotherapy for STS patients (doxorubicin and trabectedin). We evaluated drug response using an ATP-based viability assay after cell treatment with six drug doses. We used a BRCA1-mutated ovarian cancer cell line as a positive control for PARPi response. Similarly to BRCA1-mutated ovarian cancer cells, HRD^high sarcoma cells showed higher sensitivity than HRD^low sarcoma cells to PARPi but no major differential response to chemotherapy (Fig. 2a). PARPi are tested in sarcoma clinical trials in combination with standard chemotherapy, and we thus assessed potential synergism between olaparib and trabectedin in our ex vivo sarcoma cell models [5]. The combinatorial modality resulted in synergistic effects in HRD^high sarcoma cells as well as in BRCA1-mutated ovarian cancer cells but not in HRD^low sarcoma cells (Fig. 2b).

Bild vergrößern

We hypothesized that inhibitors targeting other DDR mechanisms might also exhibit HRD-dependent synthetic lethality in sarcoma cells. We focused on the serine/threonine kinase WEE1, known to inhibit CDK1 and CDK2, thus blocking G2/M progression and promoting DNA repair [10]. To analyze the effect of WEE1 inhibition on sarcoma cell viability, we subjected five HRD^high and five HRD^low patient-derived ex vivo sarcoma cell models to the WEE1 inhibitor (WEE1i) adavosertib. Adavosertib elicited a dose-dependent decrease in cell viability in HRD^high sarcoma cells (Fig. 2a). Moreover, combination of adavosertib and doxorubicin showed synergistic effects in HRD^high but not in HRD^low sarcoma cells (Fig. 2b). Altogether, our data show that targeting the DNA damage response and DNA repair pathways are powerful therapeutic strategies for sarcoma with HRDness traits.

RAD51 plays a crucial role in DNA replication and HRR, catalyzing the recognition of homology and strand exchange between partner DNA strands, one with a processed DNA break and the other acting as the repair template [7]. The formation of RAD51 nuclear foci is regarded as a functional biomarker of HRR proficiency and can be predictive of PARPi resistance [2, 3, 6]. To functionally assess HRR in sarcoma cell models ex vivo, we evaluated the formation of RAD51 nuclear foci after eliciting DNA damage with olaparib and trabectedin, either as monotherapy or in combination. Drug-induced DNA damage was evaluated by immunostaining for the phosphorylated form of the histone variant H2A.X (γH2A.X). Higher γH2A.X expression was measured in cells treated with the combination regimen compared with the single agents, regardless of their HRD status. Notably, while HRD^low sarcoma cells formed RAD51 nuclear foci upon olaparib- and trabectedin-induced DNA damage, neither formation of RAD51 nuclear foci nor increased RAD51 nuclear intensity were observed in HRD^high sarcoma cells and BRCA1-mutated ovarian cancer cells (Fig. 3a). Patient tissue samples from which the HRD^high cells were derived also showed reduced RAD51 expression and nuclear foci. Altogether, our combined results show deficiency in HRR in PARPi-sensitive patient-derived sarcoma cells.

Bild vergrößern

A 47-year-old female patient presenting with multiple lesions in the lung and liver was diagnosed in April 2020 with metastatic leiomyosarcoma, most likely of uterine origin. The patient underwent SOC and was treated with pegylated liposomal doxorubicin (PLD). Broad molecular profiling revealed TP53 loss, stable microsatellite status, tumor mutational burden of 7 mut/mb, and an LOH score of 26%, considered high in the hospital-based molecular tumor board. Due to an anthracycline-induced cardiomyopathy, PLD treatment was terminated, and the patient underwent off-label treatment with trabectedin and olaparib, remaining in a lasting radiological partial remission for one and a half years (Fig. 3b). Whole-genome sequencing of a stomach metastasis in late 2022 revealed an HRD score of 83 and genomic alterations in multiple HRR genes. Moreover, the absence of nuclear RAD51 expression in the metastatic tissue further evidenced defective HRR (Fig. 3c). This case highlights the clinical benefit of extending the therapeutic indications for PARPi alone or in combination with standard chemotherapy.

The efficacy of both olaparib and trabectedin combination therapy as well as of adavosertib in combination with chemotherapeutic agents is currently under clinical investigation in either metastatic or advanced sarcoma (NCT04076579), recurrent ovarian cancer (NCT02101775), or relapsed or refractory solid tumors in pediatric patients (NCT02095132). Results from non-biomarker- and HRD score-guided clinical trials will likely evidence the clinical benefit of DDR-targeting agents for sarcoma patients. In addition, the incorporation of transcriptomic profiling in clinical practice holds significant biomarker potential as gene expression changes faithfully reflect the rapidly changing molecular dynamics of the cell. Future research will determine whether incorporating transcriptomic biomarkers into clinical trial design can improve the clinical management of STS patients.