The dual immune landscape of homologous recombination-deficient tumors: mechanisms and therapeutic implications

Abstract

Homologous recombination deficiency (HRD) is a major genomic feature of several malignancies and provides a therapeutic rationale for DNA-damaging agents and PARP inhibitors. Beyond its role in genome instability, accumulating evidence suggests that HRD also reshapes the tumor microenvironment (TME). However, the immune consequences of HRD are not uniformly stimulatory. HRD-associated tumors may exhibit DNA damage-driven innate immune activation, increased interferon signaling, and enhanced antigen presentation, while simultaneously developing adaptive immune-evasive programs, including T-cell dysfunction, myeloid and regulatory T-cell recruitment, stromal remodeling, and vascular barriers to immune infiltration. In this review, we summarize current evidence linking HRD to immune activation and immune suppression across tumor types and propose a dual immune framework for understanding HRD-associated TME remodeling. We further discuss why HRD alone is insufficient to predict responses to immune checkpoint blockade and highlight emerging therapeutic strategies that integrate DNA damage response targeting with immune and stromal modulation.

Background

DNA double-strand break (DSB) repair is primarily achieved through four pathways: homologous recombination (HR), non-homologous end joining (NHEJ), microhomology-mediated end joining (MMEJ), and single-strand annealing (SSA) [1]. Failure to repair DSBs properly and promptly can lead to cellular transformation and the development of various diseases, including cancers [1]. HR is the only pathway capable of restoring the original DNA sequence at the damaged site [1, 2]. Defects in key HR components lead to homologous recombination deficiency (HRD)—classified as genetic, epigenetic, genomic scar–defined, or functional HRD (Table 1)—and force cells to rely on error-prone repair mechanisms, thereby promoting genomic instability and cancer risk [1, 3, 4].

Table 1.

Classification and molecular definitions of homologous recombination deficiency

Category	Definition
BRCA1/2-associated genetic HRD	Pathogenic germline or somatic BRCA1/2 alteration with biallelic tumor inactivation [2, 161]
Non-BRCA HRR gene-associated genetic HRD	Biallelic pathogenic inactivation of non-BRCA HRR genes such as PALB2, RAD51C, RAD51D, BARD1, or BRIP1 [118, 161]
Epigenetic HRD	Transcriptional silencing of HRR genes, most commonly through BRCA1 or RAD51C promoter hypermethylation [2, 7]
Genomic scar-defined HRD	A historical footprint of HR failure measured by a high HRD score/GIS (loss of heterozygosity, telomeric allelic imbalance, and large-scale state transitions), COSMIC SBS3, HRDetect, or IMPACT-HRD. Genomic scars can persist after functional HR restoration [6, 118, 161]
Functional HRD	Current inability to form RAD51 nuclear foci after DNA damage. Functional testing complements genomic assays and may identify active HRD without a canonical genetic or epigenetic driver [47]

Growing evidence suggests that HRD is a fundamental driver in the development of many malignancies, particularly breast, ovarian, and pancreatic cancers [5, 6]. Approximately 50% of high-grade serous ovarian carcinoma (HGSOC) cases exhibit an HRD phenotype due to BRCA gene mutations or BRCA promoter methylation [7]. Additionally, nearly half of pancreatic acinar cell carcinoma cases harbor germline pathogenic variants in HR or DNA damage response (DDR) genes [8]. The genomic instability of HRD tumors renders them particularly susceptible to DNA-damaging agents, and exploiting this vulnerability has become a standard therapeutic strategy. Although HRD patients often show initial sensitivity to these agents, most ultimately develop drug resistance, resulting in diminished long-term therapeutic benefit [9]. Therefore, identifying novel therapeutic targets is essential for improving clinical outcomes in HRD patients.

The tumor microenvironment (TME)—comprising tumor cells, immune cells, stromal components, and the extracellular matrix—shows substantial changes in HRD tumors [10]. Immunogenic nucleic acids comprise diverse DNA- and RNA-derived species that are detected by distinct nucleic acid-sensing pathways [11]. Cytosolic dsDNA, micronuclear DNA, and R-loop-derived RNA hybrids signal mainly through cGAS–STING, whereas transposable element-derived dsRNA is sensed by RIG-I/MDA5 [11]. On the one hand, impaired DSB repair in HRD tumors leads to the accumulation of immunogenic nucleic acids, thereby activating antitumor immunity [12–14]. On the other hand, HRD tumors employ multiple adaptive immune-evasion mechanisms to suppress immune responses [5, 15]. Thus, HRD-associated tumors should not be viewed simply as immunologically hot or immunotherapy-sensitive tumors. A major unresolved question is how HRD-driven inflammatory signaling is converted into effective immune surveillance in some contexts but into chronic immune suppression in others. In this review, we discuss HRD-associated TME remodeling through a dual immune framework that integrates tumor-intrinsic DNA damage signaling, immune-cell dysfunction, stromal remodeling, vascular exclusion, and therapeutic implications.

A dual immune framework of HRD-associated TME remodeling

HRD tumors often exhibit a dual immunological phenotype. On one side, they display heightened immune activation, characterized by increased immune infiltration and enrichment of tumor-reactive immune populations [12, 14, 16]. On the other side, HRD tumors can simultaneously develop immune-evasive mechanisms that counteract persistent DNA damage signaling and the immunoregulatory effects of HRD-associated genes [15, 17]. For example, while HRD-associated genomic instability can activate canonical cGAS–STING signaling and type I IFN-mediated antitumor immunity, persistent chromosomal instability may instead sustain STING–noncanonical NF-κB signaling and promote metastasis [11, 18]. Therefore, the HRD-associated TME is better conceptualized as a dynamic balance between DNA damage-driven immune activation and compensatory immune suppression, rather than a uniformly immune-inflamed state. This framework provides the rationale for the following sections, which first summarize immune activation driven by immunogenic nucleic acids and then examine the major mechanisms through which HRD tumors evade immune surveillance (Table 2).

Table 2.

Effects of homologous recombination deficiency on the tumor immune microenvironment

HRD definition	Tumor type	Model system	Immune readout	Direction	Evidence level	Ref
Germline BRCA1 mutation	Breast cancer	Human bulk genomic/transcriptomic cohort	Immune infiltration and immune-related gene signatures	Context-dependent	observational human	[27]
HRD-signature-high	Pancreatic cancer	Human multi-omics tumor cohort	Antigen-presenting macrophages	Immune-activating	observational human	[28]
Germline HRR gene mutations	Primary prostate cancer	Human spatial immunoprofiling cohort	T-cell infiltration and spatial distribution	Immune-activating	observational human	[29]
Genomic scar–defined HRD	HGSOC	Human single-cell tumor cohort + mouse models	T-cell and macrophage states	Mixed	observational human + mechanistic in vivo	[12]
BRCA1/2 mutations	HGSOC	Human spatial proteomic cohort	T-cell infiltration and spatial interactions	Immune-activating	observational human	[13]
BRCA1/2 mutations	HGSOC	Human multi-omics tumor cohort	Immune cell signatures and neoantigen load	Context-dependent	observational human	[14]
HRR gene mutations	HGSOC	Human multi-omics tumor cohort	T-cell infiltration	Context-dependent	observational human	[30]
HRR gene mutations	Early TNBC	Human immunogenomic cohort	CD8⁺ T-cell infiltration and PD-L1 expression	Immune-activating	observational human	[31]
BRCA1 mutations	Breast cancer	Human spatial proteomic cohort	Treg and dysfunctional T-cell spatial organization	Immunosuppressive	observational human	[15]
HRR gene mutations	Colorectal cancer	Human genomic/transcriptomic cohort	Immune infiltration and PD-L1 expression	Immune-activating	observational human	[32]
BRCA1 mutation	Breast cancer	Human immunogenomic cohorts + syngeneic mouse models	T-cell and myeloid immune states	Immunosuppressive	observational human + mechanistic in vivo	[33]
BRCA2 mutation	Breast cancer	Human immunogenomic cohorts + syngeneic mouse models	T-cell and myeloid immune states	Immune-activating	observational human + mechanistic in vivo	[33]
HRD-related genomic signatures	HER2-negative metastatic breast cancer	Human genomic/transcriptomic clinical cohort	Immune infiltration and antigen presentation	Context-dependent	prospective human	[162]
BRCA1/2 mutations	HGSOC	Human multi-omics tumor cohort	Immune signaling and TME features	Context-dependent	observational human	[17]
BRCA1 deficiency in T cells	Breast cancer	Conditional knockout mouse models	CD8⁺ T-cell function	Immunosuppressive	mechanistic in vivo	[35]
Genomic scar–defined HRD (GSS ≥ 50)	NSCLC	Human genomic/transcriptomic cohort	MHC-II, IFN-γ and CD8⁺ T-cell signatures	Immunosuppressive	observational human	[40]
BRCA1/2 mutations + HRD score	Breast cancer	Human genomic/transcriptomic cohort + IHC validation	Immune infiltration and cytolytic activity	Context-dependent	observational human	[41]
HRD score	TNBC	Human multi-scale transcriptomic cohort	T-cell and dendritic-cell states	Context-dependent	observational human	[42]
Germline or somatic mutations in BRCA1/2 or BRIP1	End-stage HGSOC	Human multi-omics tumor cohort	HLA-I loss and immune infiltration	Immunosuppressive	observational human	[43]
HRD score	Lung adenocarcinoma with brain metastasis	Human paired genomic/transcriptomic cohort	Immune infiltration and antigen presentation	Immunosuppressive	observational human	[44]
BRCA1 mutation	TNBC	Human transcriptomic cohort + BRCA1-deficient mouse models + in vitro models	IFN signaling and lymphocyte infiltration	Immunosuppressive	mechanistic in vivo + in vitro/ex vivo	[46]
BRCA1/2 mutation and/or GIS ≥ 42	HGSOC	Human single-cell/TCR cohort + mouse models	eTregs and exhausted CD8⁺ T cells	Mixed	prospective human + mechanistic in vivo	[48]
BRCA2 and ATM mutations	Metastatic castration-resistant prostate cancer	Human genomic/proteomic cohort + PDX/organoid models	B7-H3 expression	Immunosuppressive	observational human + mechanistic in vivo + in vitro/ex vivo	[53]
HRD-associated mutational signatures	HGSOC	Human multi-omics and single-cell cohort	HLA diversity and CD8⁺ T-cell states	Mixed	observational human	[54]
BAP1 loss-of-function mutations	Intrahepatic cholangiocarcinoma	Human genomic/IHC cohort + in vitro validation	B7-H4/B7-H3 expression and T-cell spatial distribution	Immunosuppressive	observational human + in vitro/ex vivo	[55]
Disrupted BRCA1–PALB2 interaction	Hepatocellular carcinoma	Genetically engineered mouse models	T-cell infiltration and PD-L1 expression	Mixed	mechanistic in vivo	[58]
BRCA1/p220 loss	Breast cancer	Human tumor cohort + syngeneic mouse models	M2 TAMs, Tregs and CD8⁺ T cells	Immunosuppressive	observational human + mechanistic in vivo	[60–62]
BRCA1 mutation	TNBC	Genetically engineered mouse model	Macrophage and T-cell infiltration	Immunosuppressive	mechanistic in vivo	[63]
BRCA1 mutation	Breast cancer	Human tumor cohort + mouse models	MDSC accumulation and CD8⁺ T-cell suppression	Immunosuppressive	observational human + mechanistic in vivo	[65]
BRCA1 mutation	Breast cancer	Genetically engineered mouse models + syngeneic tumor models	MDSCs and T-cell infiltration	Immunosuppressive	mechanistic in vivo	[66]
BAP1 mutation	Clear cell renal cell carcinoma	Human tumor cohort + mouse models	T-cell infiltration and myeloid suppression	Immunosuppressive	observational human + mechanistic in vivo	[68]
BRCA1 mutation	Breast cancer	Human breast cancer samples + mouse models	M2 TAMs, MDSCs and Tregs	Immunosuppressive	observational human + mechanistic in vivo	[69]
BRCA1/2 mutations	TNBC	Mouse breast tumor model + human CAF cohort	Immunoregulatory CAFs and CD8⁺ T-cell suppression	Context-dependent	observational human + mechanistic in vivo + in vitro/ex vivo	[73]
Germline BRCA1/2 mutations	PDAC	Human stromal transcriptomic cohort + organoids + mouse models	Immune-regulatory CLU⁺ CAFs	Immunosuppressive	observational human + mechanistic in vivo + in vitro/ex vivo	[74]
HRD-associated genomic instability	HGSOC	Human single-cell/epigenomic cohort + mouse models	POSTN⁺ myCAFs, eTregs and exhausted CD8⁺ T cells	Immunosuppressive	observational human + mechanistic in vivo	[75]
BRCA1 deficiency	Ovarian cancer	Human tumor data + genetically engineered mouse models	T-cell infiltration, type I IFN/STING signaling, VEGF-A	Mixed	observational human + mechanistic in vivo	[78]

Immunogenic nucleic acids drive immune activation in HRD-Associated TME

Due to deficiencies in HR genes such as BRCA1, BRCA2, and RAD51, HRD tumors exhibit marked genomic instability. Immunogenic nucleic acids accumulate through three major mechanisms in HRD tumors: defective DSB repair, impaired replication fork protection, and aberrant R-loop accumulation, which give rise to cytosolic dsDNA, micronuclear DNA, and RNA–DNA hybrids, respectively [1, 11, 19, 20]. Among these immunogenic nucleic acids, cytosolic dsDNA fragments and DNA released from ruptured micronuclei are sensed by cyclic GMP–AMP synthase (cGAS), whereas R-loop-derived RNA–DNA hybrids may activate both cGAS and TLR3 [11]. cGAS activation thereby initiates an innate immune response through the cGAS–STING pathway that subsequently promotes adaptive antitumor immunity [11].

During the innate immune phase, cGAS activation induces IRF3-mediated type I interferon (IFN) production, which drives a coordinated antitumor response by promoting Batf3⁺ dendritic cell (DC) maturation for tumor antigen cross-presentation, enhancing NK cell cytotoxicity, and skewing macrophages toward pro-inflammatory phenotypes [19, 21–24]. This innate activation establishes an immunogenic TME that facilitates the transition to adaptive immunity. Activated professional antigen-presenting cells (pAPCs)—notably DCs and macrophages—process and present tumor neoantigens via MHC pathways [25, 26]. In BRCA1-mutant breast cancer, DCs cross-present antigens to naive CD8⁺ T cells through elevated MHC-I [27], while HRD-associated macrophages, particularly in pancreatic cancer, contribute via MHC-II-mediated antigen presentation, thereby promoting local immune engagement [28]. The resulting activation of cytotoxic CD8⁺ T cells enables antigen-specific tumor cell elimination [29] (Fig. 1).

Fig. 1.

The activation of antitumor immunity in HRD tumors. In HRD tumors, deficiencies of HR genes lead to the accumulation of immunogenic nucleic acids through multiple mechanisms. Cytosolic immunogenic nucleic acids activate the cGAS–STING pathway, inducing type I IFN production. IFN signaling promotes innate immune cell activation, facilitates antigen presentation, and ultimately triggers CD8⁺ T-cell-mediated antitumor cell killing. HRD, homologous recombination deficiency; IFN, interferon

Consistent with these mechanisms, clinical and multiomic studies across ovarian, breast, prostate, and colorectal cancers support a broadly immune-active phenotype associated with HRD or HRR alterations, characterized by enhanced immune infiltration and T-cell-associated immune programs [12, 29–32]. Immune activation may vary across HRR genotypes, with BRCA2-deficient tumors showing stronger innate and adaptive immune features than BRCA1-deficient tumors [33].

Persistent DNA damage and the immunoregulatory roles of HRD-associated genes contribute to immune evasion in HRD tumors

Although unrepaired DNA damage can initiate immune surveillance, persistent DNA damage in HRD tumors can drive compensatory mechanisms that suppress antitumor immunity [34]. Beyond these DNA damage–associated adaptations, HRD tumors further attenuate immune responses through distinct programs [35, 36]. These effects can be broadly divided into two main pathways (Fig. 2). The first is T-cell-intrinsic, exemplified by the consequences of BRCA1 loss for T-cell development and function. The second is tumor-cell-driven remodeling of the TME and comprises four major mechanisms: direct impairment of T-cell antigen recognition and cytotoxicity; recruitment or activation of immunosuppressive immune cells; CAF remodeling; and endothelial or vascular exclusion of T cells from the tumor core.

Fig. 2.

Tumor immune evasion mechanisms driven by HRD. HRD tumors can facilitate immune evasion through two main mechanisms. First, T-cell-intrinsic BRCA1 deficiency can impair T-cell development, differentiation, memory formation, and effector function. Second, HRD tumor cells remodel the TME through four interrelated modules: (1) direct inhibition of T-cell antigen recognition and cytotoxicity; (2) recruitment and activation of immunosuppressive myeloid cells and Tregs; (3) induction of CAF states that suppress immunity; and (4) VEGF-A-associated vascular remodeling that limits T-cell entry into the tumor core. Persistent nucleic-acid sensing and aberrant transcription can contribute to these programs, converting an initially inflammatory response into chronic immune suppression. HRD, homologous recombination deficiency; TME, tumor microenvironment; IFN, interferon

Suppressing T-cell functions directly

Interfering with the development and differentiation of T-cells

T cells arise from bone marrow-derived progenitors that migrate to the thymus, where they undergo maturation before entering peripheral tissues. These tightly regulated developmental and functional programs can be disrupted in HRD tumors. Studies have found that BRCA1 deficiency can impair T-cell responses and promote the establishment of an immunosuppressive TME [35]. Biallelic deletion of BRCA1 disrupts thymocyte development, resulting in compromised T-cell differentiation [37]. At the peripheral level, mice with T-cell-specific BRCA1 deletion exhibit marked functional defects, including reduced CD8⁺ T-cell numbers, enhanced T-cell exhaustion, diminished cytotoxicity, and decreased tumor-infiltrating memory T cells [35]. Mechanistically, BRCA1 can regulate the expression of the transcription factor TCF1 [35], a transcription factor critical for sustaining CD8⁺ T-cell self-renewal and for effective antitumor immunity and responsiveness to immune checkpoint blockade [38]. Compared to wild-type mice, those with biallelic BRCA1 deletion exhibit significantly reduced TCF1 expression in CD8⁺ T cells [35]. In BRCA1 mutation carriers, reduced TCF1 expression has also been observed and is associated with increased susceptibility to breast cancer initiation [39].

Inhibiting T-cell activation by interfering with antigen presentation

Activated pAPCs—especially DCs—cross-present tumor neoantigens via MHC-I to prime naive CD8⁺ T cells, which then recognize antigen-bearing tumor cells and mediate cytotoxicity [25, 26]. In early HRD tumorigenesis, nucleic acids generated by defective DSB repair activate cGAS–STING, inducing type I IFNs that bolster pAPC activation and antigen presentation, thereby strengthening CD8⁺ T-cell immunity [24]. However, immune pressure ultimately selects for tumor cell populations that acquire mechanisms to evade antitumor surveillance.

Alterations in HLA genes represent a key mechanism of immune escape [40, 41]. In HRD breast cancer, despite the activation of DCs, HLA-I expression varies considerably across samples, suggesting heterogeneous impairment of antigen presentation capacity [42]. This heterogeneity may partly result from HRD-associated genomic instability, which can disrupt HLA-I-mediated antigen presentation through diverse genomic alterations and thereby impair T-cell recognition. Supporting this notion, multi-omic analyses in HRD HGSOC have revealed alterations affecting HLA-mediated antigen presentation, including HLA loss of heterozygosity (LOH), which may compromise immune surveillance and facilitate immune evasion during tumor evolution [43]. Similarly, HRD signatures are enriched in brain metastasis of lung cancer compared with matched primary tumors and are accompanied by impaired tumor antigen recognition and reduced immune infiltration [44].

Beyond HLA gene defects, sustained IFN signaling triggered by immunogenic nucleic acids in HRD tumors can promote adaptive mechanisms that suppress antigen presentation. MYC is critical for tumor initiation and progression in BRCA1/2-mutant breast cancer [45]. Its overexpression is frequently observed in HRD tumors and has been shown to repress the transcription of IFN signaling genes [46, 47]. Moreover, persistent IFN pathway activation can induce immunosuppressive effects. In HRD ovarian cancer, chronic IFN signaling induces a subset of tumor cells with elevated MHC-II expression, which promotes regulatory T-cell (Treg) expansion and contributes to an immunosuppressive TME [48].

Inhibiting the cytotoxic effect of effective CD8⁺ T cells

Upon recognizing tumor cells, cytotoxic CD8⁺ T cells induce apoptosis through perforin, granzymes, and Fas–FasL interactions and release cytokines such as IFNγ and TNFα, which stimulate other immune cells and suppress tumor proliferation [49]. Recent evidence indicates that ferroptosis represents an important mechanism through which CD8⁺ T cells enhance tumor cell killing [50]. However, HRD tumors may evade this T-cell-mediated cytotoxicity by enhancing antioxidant defenses that limit ferroptotic cell death. The susceptibility to ferroptosis is tightly regulated by the SLC7A11–GSH–GPX4 axis, in which SLC7A11-mediated cystine uptake supports GSH synthesis and GPX4 detoxifies lipid peroxides to prevent ferroptotic damage [51]. CD8⁺ T-cell-derived IFNγ suppresses SLC7A11 expression in ovarian cancer cells, limiting cystine uptake and GPX4-mediated antioxidant defense, thereby promoting ferroptosis [50]. Conversely, BRCA1 deficiency impairs GPX4 degradation, resulting in GPX4 accumulation and increased resistance to ferroptosis [36]. In HRD prostate cancer models, pharmacological induction of ferroptosis restrains tumor growth, increases CD8⁺ T-cell infiltration, reduces myeloid-derived suppressor cells (MDSCs), and improves response to immune checkpoint blockade [52].

T-cell function is regulated by multiple mechanisms. Among these, immune checkpoint pathways play a central role. Notably, HRD has been linked to the upregulation of immune checkpoint molecules on tumor cells, which may contribute to T-cell dysfunction [53, 54]. Compared to the tumor core, the peritumoral regions of HRD tumors exhibit higher expression of these molecules alongside an accumulation of dysfunctional CD8⁺ T cells [54, 55]. This spatial pattern suggests a mechanism by which HRD tumors impair T-cell effector function and immune surveillance. Besides, sustained activation of IFN signaling contributes to the upregulated immune checkpoint molecules. In metastatic castration-resistant prostate cancer, HRD activates the JAK–STAT–IRF1 signaling pathway—an upstream regulator of IFN responses—resulting in increased expression of B7 family members, including PD-L1 and B7-H3 [53]. Similarly, single-cell RNA sequencing has revealed that BRCA1/2 germline mutations can drive immune checkpoint upregulation in precancerous breast epithelial cells via CD8⁺ T-cell-derived IFNγ [56]. Although IFNγ–induced PD-L1 expression is not unique to HRD tumors, persistent DNA damage and chronic interferon signaling may enhance this adaptive immune resistance mechanism in HRD contexts.

Together, these mechanisms suggest that HRD-associated immune escape occurs at multiple stages of T-cell-mediated antitumor immunity, including T-cell differentiation, antigen recognition and activation, and effector cytotoxicity. The relative contribution of each mechanism likely varies by tumor type and by the specific HR gene involved, which remains incompletely understood.

Promoting the recruitment and activation of immunosuppressive immune cells

Immune cells such as macrophages, neutrophils, and regulatory T cells actively shape the tumor immune microenvironment through direct cellular interactions and diverse cytokine signals. Increasing evidence indicates that HRD tumors enhance the recruitment and activation of these immunosuppressive populations, thereby establishing a suppressive TME that supports tumor initiation and progression.

Traditionally, TAMs are classified into pro-inflammatory M1 and anti-inflammatory M2 subtypes [57]. Studies have shown that HRD can reprogram TAMs toward immunosuppressive states, including enhanced checkpoint expression and M2-like polarization. In hepatocellular carcinoma, BRCA deficiency activates cGAS–STING signaling in M1 TAMs, driving PD-L1 upregulation through the IRF3–STAT1 axis and dampening antitumor immunity [58]. In breast cancer, the BRCA1-IRIS (an alternative splice variant of the BRCA1 gene) [59] is overexpressed in BRCA1-deficient breast cancers and enhances tumor cell invasiveness through its interactions with TAMs [60–62]. BRCA1-IRIS overexpression in triple-negative breast cancer (TNBC) activates the NF-κB signaling pathway, leading to TAM recruitment and promoting their polarization toward the M2 phenotype [60]. These M2-polarized TAMs contribute to TNBC progression by promoting the epithelial-mesenchymal transition and stemness in tumor cells, thereby facilitating metastatic dissemination, while suppressing cytotoxic T-cell function and antitumor immunity [60]. In addition, enrichment of PD-1⁺ TAMs have been observed in BRCA1-mutant TNBC, and PD-1 blockade reduced these macrophages while enhancing CD8⁺ T-cell infiltration, further restoring antitumor immunity [63].

MDSCs are a heterogeneous group of neutrophils and monocytes that acquire strong immunoregulatory activity under pathological conditions [64]. In BRCA1-deficient breast tissue, epithelial cells secrete S100A9 and CXCL12, activating the S100A9–CXCL12–pSTAT3 axis in MDSCs and promoting their expansion [65]. These activated MDSCs can suppress the proliferation and function of cytotoxic T lymphocytes, thereby establishing an immunosuppressive microenvironment that facilitates tumor recurrence and distant metastasis [65]. Additionally, the loss of CUL5 has been shown to exert an even greater tumor-promoting effect in BRCA1-deficient breast cancer. Mechanistically, CUL5 deficiency in HRD tumor cells activates the CREB1–CCL2 signaling pathway, leading to increased cytokine secretion and MDSC recruitment, which ultimately contributes to reduced T-cell infiltration [66]. These observations suggest a cooperative role for BRCA1 and CUL5 in shaping the TME, in which BRCA1 loss may downregulate CUL5, thereby amplifying MDSC recruitment and reinforcing immunosuppression.

Tregs exert immunosuppressive effects that attenuate antitumor immunity [67]. In HRD ovarian cancer, sustained activation of the cGAS–STING pathway induces the expression of MHC class II molecules on tumor cells, which in turn promotes Treg differentiation through the upregulation of co-inhibitory signaling molecules [48]. In addition, BAP1-mutant clear cell renal cell carcinoma (ccRCC) enhances the recruitment of CCR5⁺ Tregs into the TME by significantly upregulating CCR5 and its ligands, further reinforcing an immunosuppressive TME [68].

In addition to mechanisms that target the activation of specific pro-tumoral immune cells, VEGFA has been shown to promote aggressive tumor phenotypes and broadly facilitate the enrichment of immunosuppressive immune cells. In BRCA1-mutant breast cancers, upregulation of VEGFA and IL-6 promotes the accumulation of polysialic acid on breast epithelial cells through activation of the TGFβ–ST8SIA4 signaling pathway [69]. Polysialic acid enrichment reduces epithelial cell adhesion and enhances migratory and metastatic capabilities, while simultaneously shaping an acidic tumor-permissive microenvironment characterized by the enrichment of immunosuppressive cells—including M2 TAMs, MDSCs, and Tregs [69].

Collectively, these studies indicate that myeloid and regulatory immune populations are not merely secondary responders to HRD-associated inflammation, but may actively determine whether HRD tumors remain immune-inflamed or progress toward immune suppression. Future studies should distinguish whether these populations are induced by DNA damage signaling itself, by specific HR-gene-dependent transcriptional programs, or by treatment-induced selective pressure.

Interactions with CAFs to facilitate immune evasion

CAFs display considerable functional heterogeneity [70] and are broadly categorized into three subtypes: myofibroblastic CAFs (myCAFs), which drive ECM production; inflammatory CAFs (iCAFs), which secrete immunoregulatory cytokines; and antigen-presenting CAFs (apCAFs), which exhibit antigen-presenting capacity [71, 72]. A higher proportion of iCAFs in BRCA1-mutated tumors is associated with a significant reduction in immunosuppressive cell populations [42]. In contrast, myCAFs in non-HRD tumors exhibit a distinct DPP4 + phenotype, which is closely linked to decreased immune cell recruitment [42]. Similarly, PDPN⁺ pCAFs are reduced in BRCA1/2-mutant TNBC, and an immunoregulatory subset of these cells can suppress CD8⁺ T-cell activation and proliferation [73].

Although immune-supportive CAF states have been identified, sequencing studies reveal pronounced CAF heterogeneity, including immunosuppressive subsets. In BRCA mutant pancreatic ductal adenocarcinoma (PDAC), cancer cells secrete regenerating islet-derived proteins that activate HSF1 signaling in pancreatic stellate cells, driving their differentiation into CLU⁺ CAFs, which suppress T-cell activation [74]. A recent publication also revealed the ability of HRD tumors to remodel CAF states. In HRD HGSOC, genomic instability activates tumor-cell STING signaling, which induces Wnt-dependent differentiation of fibroblasts into POSTN⁺ myCAFs [75]. These POSTN⁺ myCAFs promote Treg enrichment and CD8⁺ T-cell exhaustion, ultimately establishing an immunosuppressive TME [75].

These findings highlight the context-dependent roles of CAFs in HRD-associated tumors. Some CAF subsets may be reduced in HRD tumors or may correlate with a less suppressive immune state, whereas other fibroblast programs can promote immune exclusion. Therefore, CAF heterogeneity should be incorporated into future HRD-based immunotherapy stratification rather than treated as a uniform stromal feature.

Utilizing endothelial cells to restrict T-cell infiltration into the tumor core

Endothelial cells regulate chemokine secretion and adhesion molecule expression to support T-cell trafficking into tumor cores, whereas elevated VEGF drives abnormal angiogenesis and impairs effective T-cell infiltration [76].

VEGFA expression is markedly higher in HRD tumors than in HR-proficient (HRP) tumors [77, 78], and several HRD-associated mechanisms contribute to this increase. One key mechanism is that BRCA1 suppresses VEGFA transcription by interfering with estrogen signaling [79, 80]. Loss of BRCA1 relieves its transcriptional repression of VEGFA, resulting in enhanced angiogenesis in HRD tumors. Sustained activation of the STING pathway in BRCA1-mutant tumors may also contribute to VEGFA regulation [78, 81]. Elevated VEGFA expression promotes tumor angiogenesis and suppresses the infiltration of CD8⁺ tumor-infiltrating lymphocytes [78].

Overall, vascular remodeling provides a plausible explanation for why HRD tumors with inflammatory signaling may still show poor intratumoral T-cell penetration. This supports the rationale for combining DNA damage response targeting with anti-angiogenic or vascular-normalizing strategies in selected HRD tumors.

Therapeutic implications: exploiting and overcoming the dual immune state of HRD tumors

HRD tumors create both therapeutic opportunities and challenges. Their defects in homologous recombination can heighten tumor immunogenicity, yet the same tumors may also engage diverse mechanisms that dampen immune activation. These dual properties indicate that effective therapeutic strategies should not merely intensify DNA damage, but should also overcome tumor-intrinsic, immune-cell, stromal, and vascular barriers that limit productive antitumor immunity. Accordingly, current approaches can be organized into three major directions: (1) DNA-damaging agents that enhance immunogenic nucleic acid accumulation; (2) immunotherapies or microenvironment-directed therapies that reverse immune suppression; and (3) rational combination or sequencing strategies that integrate both principles to maximize therapeutic efficacy (Fig. 3).

Fig. 3.

Therapeutic strategies targeting the dual immune state of HRD tumors. Three immune-related therapeutic approaches are shown: DNA-damaging agents that induce immunogenic nucleic acids, therapies targeting the suppressive TME, and combination strategies that integrate DNA damage response targeting with immune or stromal modulation. HRD, homologous recombination deficiency; TME, tumor microenvironment; PARPi, PARP inhibitor; ICB, immune checkpoint blockade; NA, nucleic acids

Inducing DNA damage to enhance the accumulation of immunogenic nucleic acids in HRD tumors

Synthetic lethality refers to a phenomenon in which a defect in either of two genes alone has no significant impact on cell viability, but the simultaneous loss of both results in cell death [82]. PARP inhibitors exploit synthetic lethality in HRD tumors by trapping PARP at DNA break sites and disrupting its repair catalytic cycle, thereby increasing the DNA repair burden on HRD tumor cells and enabling precision therapy [83]. However, the development of PARPi resistance limits their efficacy. Studies have shown that the efficacy of PARP inhibitors in HRD tumors depends on the activation of the cGAS–STING pathway, which promotes antitumor immunity [84, 85]. Consistently, pharmacological STING activation combined with PARP inhibition enhanced antitumor immunity and further suppressed tumor growth in immunocompetent syngeneic mouse models of BRCA1-deficient breast cancer [86]. In this context, the accumulation of immunogenic nucleic acids plays a pivotal role, acting as a critical component of PARPi-induced antitumor immunity. Combining PARPi with other DNA-damaging agents that promote the production of immunogenic nucleic acids holds promise for enhancing treatment efficacy. Several mechanisms have been discovered for the accumulation of immunogenic nucleic acids, including the suppression of HR-related gene expression, disruption of the HR process, targeting alternative DNA repair pathways, and interference with cell cycle progression (Table 3).

Table 3.

Therapeutic strategies exploiting or inducing HR deficiency to enhance immunogenic nucleic acid accumulation

Target	Drugs	HRP/HRD	Tumor type	Mechanisms	Type of research/outcome	Ref
PRMT5	PRMT5 inhibitors (GSK3326595 and JNJ64619178)	HRP	Cholangiocarcinoma model	Suppress HR-related gene expression, impair DNA repair, and activate innate immune signaling	Preclinical/positive	[90]
	GSK3326595 combined with niraparib/talazoparib	Both	Breast and ovarian cancer models		Preclinical/positive	[163, 164]
	PRMT5 inhibitors (LLY-283) combined with olaparib	Both	Breast cancer model		Preclinical/positive	[92, 93]
Intracellular PD-L1	PD-L1 depletion combined with olaparib	HRP	Cancer cell lines + syngeneic mouse tumor models	Induce BRCAness through PD-L1–BARD1–BRCA1 disruption and activate innate immunity	Preclinical/positive	[96]
Guanosine diphosphate (GDP)-mannose (GDP-M),	GDP-M supplementation combined with platinum/PARPi	low HRD score	Mouse TNBC model	Promote ubiquitin-mediated BRCA2 degradation, impair HRR, and activate innate immunity	Preclinical/positive	[99]
SETD2	DNA hypomethylating agents, HMA, combined with talazoparib	SETD2-deficient/HRD-like	Human ccRCC models	Impair BRCA1/RAD51 recruitment to DNA damage sites and activate innate immunity	Preclinical/positive	[107]
NR1D1	NR1D1 agonist (SR9009)	HRP	Mouse breast cancer model	Induce a BRCAness phenotype by suppressing HR repair and activate the innate immunity	Preclinical/positive	[110]
POLQ/POLθ	POLQ inhibitors (ART558 and novobiocin) combined with PARPi	HRD	Human and murine PDAC models, breast and ovarian tumor models	Exploit HRD-associated synthetic lethality and activate cGAS–STING signaling	Preclinical/positive	[112, 113]
	POLQ inhibitors (ART6043 and novobiocin) combined with olaparib	HRD	Advanced/metastatic solid tumors		Phase I/IIa/recruiting	NCT05898399
	POLQ inhibitors (GSK4524101) combined with niraparib	gBRCAmut	Advanced/metastatic solid tumors		Phase I/II/active, not recruiting	NCT06077877
	POLQ inhibitors (ART4215)	Both	Advanced/metastatic solid tumors		Phase I/completed	NCT04991480
CDC7	CDC7 inhibitor (XL413) combined with PARPi	HRD	PARPi-resistant ovarian cancer model	Enhance DNA damage and replication stress and activate cGAS–STING signaling	Preclinical/positive	[124]
ATR	ATR inhibitor (AZD6738) combined with olaparib	Both	TNBC xenograft model	Suppress recombination-mediated DNA repair and activate cGAS–STING signaling	Preclinical/positive	[125]
			TNBC, ovarian cancer, and prostate cancer		Phase II;/active, not recruiting	NCT03330847, NCT03462342, and NCT03787680
	Ceralasertib	DDR-deficient	Advanced solid tumors	Enhance replication-associated DNA damage and inflammatory immune signaling	Phase I/durable responses in selected DDR-defective tumors	[126]
CHK1	Prexasertib combined with olaparib	BRCA-mutant	Ovarian cancer		Phase I/completed, no results	NCT03057145
WEE1	WEE1 inhibitor (adavosertib) combined with olaparib	HRP	TNBC model	Increase DNA damage, apoptosis, replication stress, and STING pathway activation	Preclinical/positive	[128]
		Both	TNBC		Phase II/active, not recruiting	NCT03330847

Suppression of HR-related gene expression

Epigenetic regulation plays a crucial role in modulating DNA repair mechanisms by influencing the expression of HR-related genes [87]. Protein arginine methyltransferase 5 (PRMT5) regulates transcription and DNA damage responses through arginine methylation and is frequently overexpressed in multiple cancers [88–91]. In cholangiocarcinoma, PRMT5 inhibition suppresses the expression of HR genes such as BRCA1 and BRCA2, thereby leading to impaired DNA repair [90]. Beyond its effects on DNA repair, PRMT5 inhibition has been shown to activate innate immune pathways within tumor cells, enhancing the recruitment of T cells to the tumor microenvironment, thus promoting antitumor immunity [92]. Preclinical studies have also demonstrated that combining PRMT5 inhibitors with PARPi increases DNA damage and enhances antitumor immune responses in breast and ovarian cancers [92, 93].

In addition to transcriptional regulation of HR-related genes, post-transcriptional mechanisms may also influence HRD and PARPi sensitivity. Beyond its canonical role as a membrane-bound immune checkpoint ligand, intracellular PD-L1 can function as an RNA-binding protein that stabilizes HR repair transcripts, including NBS1 and BRCA1, thereby promoting resistance to DNA-damaging therapies [94, 95]. One mechanistic study showed that tumor-intrinsic PD-L1 interacts with BARD1 to promote BRCA1 nuclear accumulation and HR repair while suppressing cGAS–STING signaling [96]. In preclinical tumor models, PD-L1 depletion enhanced TBK1/CCL5 signaling and restored PARPi sensitivity [96]. These findings suggest that targeting intracellular PD-L1 may exert antitumor effects by impairing DNA damage repair and enhancing innate immune signaling.

Furthermore, accumulated metabolic byproducts in tumor cells play a crucial role in regulating the homeostasis of HR-related proteins and in activating antitumor immunity. Mutations in metabolic enzymes in malignant tumors can lead to the accumulation of 2-hydroxyglutarate (2-HG), succinate, and fumarate, all of which have been shown to inhibit homologous recombination repair [97, 98]. A recent study identified guanosine diphosphate-mannose (GDP-M) as an endogenous metabolite that promotes BRCA2 degradation and suppresses HR repair [99]. In TNBC cells, GDP-M accumulation disrupts the interaction between BRCA2 and ubiquitin-specific peptidase 21 (USP21), leading to BRCA2 degradation via a ubiquitin-mediated pathway, ultimately impairing HR repair [99]. Notably, the combination of GDP-M with PARP inhibitors has been demonstrated to activate STING-dependent antitumor immunity in murine models, significantly enhancing the efficacy of anti-PD-1 therapy [99].

Disruption of the homologous recombination process

Epigenetic regulation can impair the HR repair process, leading to the accumulation of immunogenic nucleic acids. AT-rich interaction domain protein 1 A (ARID1A), a key subunit of the SWI/SNF chromatin remodeling complex, is frequently mutated in multiple solid tumors [100, 101]. Studies have shown that ARID1A accumulates at sites of DSBs following DNA damage, where it regulates chromatin accessibility and recruits repair factors to facilitate HR repair [102]. Loss of ARID1A results in the accumulation of DNA damage and activation of the cGAS–STING signaling pathway, contributing to increased immune cell infiltration in tumors [102]. SETD2 is the primary methyltransferase responsible for catalyzing histone H3 lysine 36 trimethylation (H3K36me3) and is frequently mutated in various human cancers [103, 104]. H3K36me3 facilitates HR repair by recruiting key repair factors such as RAD51 and BRCA1 to DSB sites, thereby ensuring efficient DNA repair and maintaining genomic stability [105, 106]. Studies in renal cell carcinoma have shown that SETD2 mutations lead to the loss of H3K36me3, resulting in impaired DNA repair and the accumulation of DNA damage [107]. This damage can activate the cGAS–STING pathway, triggering the expression of type I IFNs and related immune factors [107]. Furthermore, H3K36me3 deficiency induces the expression of transposable elements, mimicking viral infection and amplifying immune responses [107]. Clinically, ARID1A and SETD2 alterations are being explored as biomarkers of DNA repair vulnerability for patient stratification in trials of DDR-targeted therapies.

Additionally, the nuclear receptor NR1D1, a core component of the biological clock, has been implicated in the regulation of DNA damage repair and antitumor immunity [108]. A study showed that NR1D1 binds to DSB sites and limits the recruitment of SIRT6, pNBS1, and BRCA1, thereby suppressing HR repair in breast cancer cells [109]. Subsequent work further suggested that NR1D1 promotes cytosolic DNA accumulation and cGAS–STING activation, leading to increased type I IFNs, CCL5, and CXCL10 [110]. In NR1D1-deficient mouse models, type I IFN expression is significantly reduced, accompanied by a marked decrease in intratumoral infiltration of CD8⁺ T-cells and NK cells [110].

Targeting alternative DNA repair pathways

MMEJ regulated by polymerase θ (POLQ) has been identified as the major compensatory repair pathway in HRD cancers [111]. Studies have shown that HR pathway inactivation is often accompanied by POLQ overexpression, which suppresses RAD51-mediated HR repair while promoting the error-prone MMEJ pathway [111, 112]. POLQ knockdown leads to the accumulation of cytoplasmic micronuclei, which activates the cGAS–STING signaling pathway and enhances CD8⁺ T-cell infiltration in BRCA2-deficient pancreatic tumors [112]. Notably, a specific POLQ inhibitor, novobiocin (NVB), selectively kills HRD tumor cells and effectively overcomes PARPi resistance, either as a monotherapy or in combination with PARPi [113]. Based on these preclinical findings, POLQ inhibitors in combination with PARPi are now being evaluated in early-phase clinical trials in patients with HRD tumors (NCT05898399, NCT06077877, NCT04991480).

Interference with cell cycle progression

BRCA2-mutated metastatic prostate cancers exhibit upregulation of cell cycle genes and enrichment of cell cycle signaling programs [114], suggesting the close association between HRD and cell cycle dysregulation. Retinoblastoma protein (RB1) is a key tumor suppressor that prevents cell cycle progression into the S phase in the presence of unrepaired DNA damage [115]. Additionally, RB1 interacts with BRG1 and ATM to facilitate HR repair [115]. Loss of RB1 expression is observed in approximately 20% of HGSOC and 40% of TNBC cases [116, 117]. Pan-cancer analyses indicate that RB1 deficiency is associated with a high genome-wide loss of heterozygosity score, a hallmark of genomic scarring indicative of HRD [118]. In BRCA1/2-deficient HGSOC, RB1 loss leads to dysregulated cell cycle control and sustained DNA damage accumulation [119]. Concurrent RB1 loss and BRCA deficiency are associated with enhanced cGAS–STING signaling and increased CD8⁺ T-cell infiltration, together with reduced EMT-related signaling linked to metastatic progression [119]. Analyses of patient samples have demonstrated that RB1–BRCA co-mutation is strongly associated with improved immunologic response and overall survival in patients with ovarian and breast cancer [14, 119–122]. Rather than representing a direct therapeutic target, RB1 loss may serve as a biomarker for stratifying BRCA-deficient tumors with distinct immune phenotypes and clinical outcomes.

HRD tumors experience heightened replication stress, making them particularly sensitive to the inhibition of replication checkpoint kinases. CDC7, a serine-threonine kinase frequently upregulated in tumors, further exacerbates replication stress, leading to increased DNA damage and the accumulation of immunogenic nucleic acids [123]. Inhibiting CDC7, especially in combination with PARP inhibitors, significantly enhances cGAS–STING pathway activation, improves antigen presentation, and strengthens interferon-mediated antitumor immunity in BRCA2-deficient advanced ovarian cancer models [124]. Similarly, the ATR inhibitor ceralasertib enhances the effectiveness of olaparib in both HRD and HRP TNBC by activating ATM-dependent signaling and disrupting homology-directed DNA repair [125]. A phase I study of ceralasertib in advanced solid tumors reported durable responses in selected tumors with genomic and DDR defects, including HRD-associated genomic instability, with responding tumors showing a more inflamed immune phenotype [126]. WEE1 inhibitors can bypass the limitation of genetic background, making BRCA1/2 wild-type TNBC more responsive to PARP inhibition [127, 128]. This synergy is primarily driven by increased DNA damage and activation of the cGAS–STING pathway, ultimately enhancing the antitumor response [128]. These findings have led to the clinical exploration of PARPi combined with replication checkpoint inhibitors (NCT03330847, NCT03462342, NCT03787680, NCT03057145, and NCT03330847).

Targeting the immunosuppressive TME caused by impaired adaptive immunity

Multiple mechanisms collectively shape the suppressive immunophenotypes of HRD tumors, ultimately driving tumor progression. Targeting these mechanisms may not only inhibit tumor growth but also help overcome therapeutic challenges associated with HRD tumors. ICB plays a crucial role in cancer immunotherapy [129]. Currently, only a limited number of biomarkers, including tumor mutation burden (TMB), PD-L1 expression, and DNA mismatch repair deficiency, have been established as predictors of response to ICB [130]. Given the unique genomic characteristics and TME of HRD tumors, HRD itself has emerged as a potential biomarker for predicting ICB sensitivity. Evidence from clinical trials suggests that HRD or HRD-associated genomic features are associated with improved responses to ICB-containing therapy in NSCLC and HGSOC [131, 132]. Clinical biomarker analyses in advanced urothelial carcinoma further link DDR alterations, including HR-related alterations, to greater benefit from PD-1/PD-L1 blockade [133, 134]. In the context of combination immunotherapy, dual ICB (CTLA-4 inhibitor and PD-1/PD-L1 inhibitor) has shown promising efficacy in HRD patients. A study investigating the use of ipilimumab plus nivolumab in patients with metastatic pancreatic or biliary tract cancer harboring pathogenic germline HRD mutations demonstrated a significant association between germline HRD status and response to ICB [135]. Similarly, in patients with BRCA-mutant castration-resistant prostate cancer, the combination of ipilimumab and nivolumab has been observed to elicit a moderate antitumor response [136].

However, not all HRD tumors are responsive to ICB therapy. BRCA1-mutated murine HGSOC models show little response to PD-L1 blockade [137]. A randomized, double-blind trial evaluating the efficacy of atezolizumab in ovarian cancer (NCT03038100) revealed that neither BRCA1/2 mutations nor HRD significantly enhanced the therapeutic benefit of atezolizumab [138]. This finding suggests that genomic instability induced by BRCA1/2 mutations or HRD does not necessarily confer increased sensitivity to ICB therapy.

Differences in cancer type and ICB regimen (monotherapy vs. dual checkpoint blockade) may influence the efficacy of ICB in HRD tumors. The discrepancy in treatment response has driven further investigation into predictive biomarkers of ICB efficacy in HRD tumors. A cohort study indicated that the combination of BRCA2 alterations and TMB may serve as a potential predictor of response to ICB [139]. Additionally, an analysis of gene expression and clinical data from over 1000 breast cancer patients revealed significant upregulation of CXCL10, a downstream target of the cGAS–STING pathway, in HRD tumors [140]. CXCL10 expression correlates positively with neoantigen load and immune cell infiltration, and may serve as a complementary biomarker to HRD for predicting response to PD-1/PD-L1 blockade [140]. Moreover, IL-6 has been identified as a key modulator in the ICB treatment of HRD tumors [141]. Dual blockade of IL-6 and immune checkpoint pathways significantly prolongs overall survival in murine models of BRCA2-mutant pancreatic cancer [141]. Mechanistically, IL-6 inhibition alleviates the immunosuppressive functions of MDSCs and macrophages, thereby restoring Th1 immune responses, enhancing T-cell infiltration into the tumor, and ultimately increasing sensitivity to ICB therapy [141].

Combining DNA damage therapy with immune or stromal modulation

Research has shown that PARP inhibitors exert immunoregulatory effects through multiple mechanisms, providing a theoretical basis for combining PARPi with immunotherapy. First, PARPi promotes the accumulation of cytosolic DNA fragments, which activate the cGAS–STING pathway in HRD cells, leading to the production of type I interferons [142, 143]. Second, PARPi can activate paracrine signaling in DCs, synergizing with the cGAS–STING pathway to recruit CD8⁺ T cells [84]. Additionally, in the neoadjuvant therapy of BRCA-mutant breast cancer, talazoparib has been shown to significantly increase the density of cytotoxic CD8⁺ T cells in both the tumor and stromal compartments by modulating macrophage–T-cell interactions [144]. These findings reveal key mechanisms through which PARPi modulates the immune microenvironment in HRD tumors.

PARPi combined with ICB

The combination of PARPi and ICB is one of the most extensively studied therapeutic strategies for HRD tumors. Preclinical studies have shown that the combination of olaparib and PD-L1 blockade significantly prolongs survival of BRCA1-deficient ovarian cancer mouse models [137]. Furthermore, combining olaparib with PD-1 blockade further enhanced STING-dependent antitumor immunity and improved tumor control compared with olaparib monotherapy in BRCA1-deficient murine ovarian cancer [85].

Consistent with these findings, early-phase clinical studies have reported encouraging antitumor activity of PARPi–ICB combinations across several solid tumors harboring BRCA1/2 alterations [145–149]. Notably, the therapeutic response to this combination is influenced by specific tumor and molecular characteristics. For instance, BRCA1/2-mutant tumors with high TMB exhibit higher objective response rates to talazoparib plus avelumab [145]. Similarly, niraparib plus dostarlimab has demonstrated potential antitumor activity in pleural mesothelioma patients harboring germline BAP1 and/or somatic BRCA2 mutations, whereas limited efficacy was observed in patients with germline BAP1 mutations alone [150]. Differential efficacy of PARPi–ICB combinations across HRD genotypes has also been reported in metastatic pancreatic cancer, with patients harboring BRCA2 or PALB2 mutations showing longer PFS and OS than those with BRCA1 mutations [149].

Despite these encouraging early clinical findings, randomized trials have not consistently demonstrated an additional benefit from incorporating ICB into PARPi-based therapy. In the ANITA trial, the addition of atezolizumab to platinum-based chemotherapy followed by maintenance of niraparib failed to significantly improve progression-free survival (PFS) or objective response rate in patients with BRCA-mutated ovarian cancer [151]. Similarly, in FIRST, adding dostarlimab to chemotherapy and niraparib maintenance produced only a modest PFS improvement in the overall population, with no clear PFS benefit observed in the HRD subgroup [152]. In the ATHENA-COMBO trial, adding nivolumab to rucaparib did not improve PFS compared with rucaparib monotherapy, including among patients with HRD tumors [153]. By contrast, in DUO-O, durvalumab plus olaparib within a bevacizumab-containing regimen improved PFS versus bevacizumab-based control in patients with HRD-positive, non-tBRCAm ovarian cancer [154]. However, because DUO-O evaluated a multicomponent regimen without a PARPi-alone comparator, the incremental contribution of ICB to PARP inhibition could not be determined. Collectively, these trials suggest that increasing HRD-associated DNA damage with PARP inhibition does not necessarily improve the clinical response to ICB. This apparent disconnect between the immunogenic consequences of HRD observed in early-phase clinical studies and the limited benefit of PARPi–ICB combinations in randomized trials suggests that HRD-associated immune activation may be counterbalanced by tumor-intrinsic and microenvironmental immunosuppressive mechanisms.

Currently, multiple clinical trials are evaluating the safety and efficacy of PARPi–ICB combination therapy in HRD cancers (NCT04985721, NCT04123366, NCT04666740, NCT03025035, NCT04417192, NCT04940637, NCT04493060, NCT04584255, NCT05659914, NCT06419179, NCT02571725, NCT02953457). These studies are expected to provide deeper clinical insights and optimized therapeutic strategies for HRD tumors.

PARPi combined with anti-angiogenic therapy

Beyond PARPi–ICB combinations, PARPi combined with anti-angiogenic therapy has also shown promising efficacy in HRD tumors. In BRCA1-deficient murine ovarian cancer, adding VEGF-A blockade to PARPi–ICB produced the greatest tumor regression compared with monotherapy or PARPi–ICB, accompanied by increased CD8⁺ T-cell infiltration and enhanced effector function [78]. The phase III PAOLA-1/ENGOT-ov25 trial demonstrated that olaparib plus bevacizumab significantly improved progression-free survival in patients with HRD-positive ovarian cancer, with long-term follow-up further showing a clinically meaningful overall survival benefit [155, 156]. Besides, another phase III clinical trial (NCT06580314) is underway to further evaluate the efficacy and safety of this combination in HRD ovarian cancer. Beyond ovarian cancer, the therapeutic value of PARPi–anti-angiogenic combinations is also being actively explored in other HRD tumors. Several ongoing clinical trials are investigating their potential applications in breast cancer (NCT06255392, NCT04090567) and small cell lung cancer (NCT04939662).

PARPi combined with monoclonal antibodies

Given the unique immune microenvironment of HRD tumors, combining PARPi with monoclonal antibodies targeting myeloid cells has shown significant therapeutic potential in preclinical studies, primarily in BRCA1/2-associated breast cancer models. Macrophage-mediated immunosuppression has been identified as a key factor limiting PARPi efficacy. Preclinical evidence suggests that PARPi can reshape macrophage function by reprogramming glucose and lipid metabolism via the sterol regulatory element-binding protein 1 (SREBP1) pathway [157]. Anti-CSF-1R antibodies target TAMs by blocking CSF-1-dependent survival signals [158]. In a BRCA1-deficient TNBC model, the combination of PARPi and CSF-1R blockade enhanced both innate and adaptive antitumor immune responses and significantly prolonged survival [157]. Consistently, macrophage depletion through CSF-1R blockade enhanced PARPi efficacy in BRCA1-deficient breast cancer models [86]. Building on these promising preclinical findings, a phase I clinical trial (NCT06488378) is currently assessing the safety and efficacy of olaparib combined with the CSF-1R-blocking antibody axatilimab in BRCA1/2 and PALB2-associated metastatic breast cancer.

PARPi combined with CAF-targeted therapy

Preclinical studies have shown the therapeutic potential of combining PARPi with CAF-targeted strategies in HRD tumors, particularly in ovarian cancer. PARPi-induced CAF remodeling can contribute to therapeutic resistance. Specifically, CCL5 neutralization attenuated PARPi-induced CAF activation and enhanced PARPi efficacy in BRCA1/2-mutant ovarian cancer xenograft models [159]. Similarly, suppression of PARPi-induced CAF senescence with bepotastine restored PARPi sensitivity in HRD-positive ovarian cancer PDX models [160]. In addition, CAF remodeling driven by HRD-associated genomic instability can promote PARPi resistance. A recent study identified an immunosuppressive POSTN⁺ myCAF population that contributed to this resistance, while targeting these cells restored T-cell cytotoxicity and enhanced PARPi efficacy in BRCA1/2-deficient ovarian and breast cancer models [75].

Toward composite biomarkers and rational trial design

The translational challenge is to move beyond the assumption that HRD status alone is sufficient to define immunotherapy sensitivity. HRD should be integrated with functional immune features, including TMB, antigen-presentation integrity, IFN activity, CXCL10 expression, T-cell infiltration, T-cell exhaustion, myeloid/Treg enrichment, CAF signatures, vascular exclusion, and prior exposure to PARPi or platinum agents. Such multidimensional stratification may help explain why some HRD tumors respond to ICB or PARPi-based combinations, whereas others remain resistant despite genomic evidence of DNA repair deficiency.

Future clinical trials should therefore consider biomarker-enriched designs that distinguish inflammatory HRD tumors from immune-excluded or myeloid/stromal-dominant HRD tumors. In inflammatory HRD tumors with intact antigen presentation and active T-cell infiltration, PARPi-ICB combinations may be most rational. In immune-excluded HRD tumors with VEGF-driven vascular barriers, PARPi plus anti-angiogenic therapy may be prioritized. In myeloid- or CAF-dominant HRD tumors, macrophage-targeting, IL-6 blockade, CSF-1R inhibition, or stromal-modulating strategies may be required to convert DNA damage-driven inflammation into productive antitumor immunity.

Conclusions and perspectives

Tumors with HRD exhibit a dual immune phenotype. Defects in homologous recombination repair can trigger immune activation through cGAS–STING signaling, type-I IFN production, antigen presentation, and CD8⁺ T-cell recruitment. However, HRD-associated immune activation is frequently counterbalanced by adaptive immune-evasion mechanisms involving T-cell dysfunction, suppressive myeloid and regulatory immune populations, CAF remodeling, and vascular exclusion. Therefore, HRD tumors should not be viewed simply as immune-hot tumors; rather, they represent a dynamic immune state in which DNA damage-driven inflammation and compensatory immune suppression coexist.

Several key questions remain to be addressed. First, it is important to clarify when HRD-associated immune stimulation gives way to a suppressive immune state and how this transition depends on specific HR gene defects, tumor type, disease stage, and treatment exposure. Second, different routes of immunogenic nucleic acid generation may not produce equivalent immune outcomes, and future studies should determine whether these routes preferentially induce immune activation, exhaustion, tolerance, or stromal exclusion. Third, clinical translation will require composite biomarkers that integrate HRD status with antigen-presentation integrity, IFN activity, T-cell infiltration and exhaustion, myeloid/Treg suppression, CAF programs, vascular exclusion, and prior PARPi exposure. Such integrated stratification may better identify which HRD tumors require PARPi, immune checkpoint blockade, anti-angiogenic therapy, myeloid targeting, or rational sequencing of these modalities. Addressing these questions will be essential for converting the immune effects of HRD into durable therapeutic benefit.

Abbreviations

APC

Antigen-presenting cell

CAF

Cancer-associated fibroblast

ccRCC

Clear cell renal cell carcinoma

cGAS

Cyclic GMP–AMP synthase

DC

Dendritic cell

DDR

DNA damage response

DSB

DNA double-strand break

dsRNA

Double-stranded RNA

ECM

Extracellular matrix

GIS

Genomic instability score

HGSOC

High-grade serous ovarian carcinoma

HR

Homologous recombination

HRD

Homologous recombination deficiency

HRP

Homologous recombination-proficient

HLA

Human leukocyte antigen

ICB

Immune checkpoint blockade

IFN

Interferon

LOH

Loss of heterozygosity

MHC

Major histocompatibility complex

MMEJ

Microhomology-mediated end joining

MDSC

Myeloid-derived suppressor cell

NK cell

Natural killer cell

NHEJ

Non-homologous end joining

NSCLC

Non-small cell lung cancer

OS

Overall survival

PDAC

Pancreatic ductal adenocarcinoma

PARPi

PARP inhibitor

PDX

Patient-derived xenograft

PFS

Progression-free survival

pAPC

Professional antigen-presenting cell

Treg

Regulatory T cell

SSA

Single-strand annealing

STING

Stimulator of interferon genes

TNBC

Triple-negative breast cancer

TAM

Tumor-associated macrophage

TME

Tumor microenvironment

TMB

Tumor mutation burden

VEGF-A

Vascular endothelial growth factor A

Author contribution

YY and XW conceived the study. YY performed data curation and investigation and drafted the manuscript. YJY contributed to conceptualization and supervision and revised the manuscript. SR participated in the investigation and manuscript revision. YQ and DL contributed to the investigation. YZ and XW acquired funding. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grants U24A20739 and 82373250) and the Program of Shanghai Academic/Technology Research Leader (Grant 22XD1402000).

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Ethical approval

Not applicable.

Consent for publication

The authors confirm their consent to publish the manuscript.

Competing interests

The authors declare no competing interests.