Deep inside every solid tumor, a quiet struggle plays out between malignant cells and the immune system — and a sweeping new study reports that this struggle unfolds differently depending on whether the patient is male or female. In research published on August 26, 2026, in the journal Biology of Sex Differences, a team at West China Hospital of Sichuan University assembled a pan-cancer single-cell atlas of unusual scale, integrating 1,662 tumor samples drawn from fourteen human cancer types. Led by co-first authors Huancheng Fu, Yajiao He and Yiqi Deng, with Heng Xu and Yang Shu as corresponding authors, the analysis revealed that biological sex leaves a pronounced imprint on the tumor microenvironment, the cellular ecosystem that surrounds and shapes a growing cancer. Female tumors carried significantly higher proportions of three cell types with well-documented immunosuppressive credentials: myofibroblastic cancer-associated fibroblasts, SPP1-positive macrophages, and exhausted CD4-positive T cells. In parallel mouse experiments, anti-PD-1 immunotherapy — the backbone of modern cancer treatment — proved significantly more effective in males than in females, while deleting a single macrophage gene, Spp1, made female mice responsive to a therapy that had been leaving them behind.

The timing of the finding is no accident. Drugs that block PD-1, a receptor that tumors co-opt to disengage patrolling T cells, have reshaped oncology over the past decade; the underlying discovery earned James Allison and Tasuku Honjo the 2018 Nobel Prize in Physiology or Medicine, and checkpoint inhibitors are now approved across dozens of cancer indications. Yet even in diseases where they shine, durable responses occur in only a minority of patients, and oncologists still lack dependable tools to predict who will benefit. Sex has rarely featured in that calculation. Biomedical research has historically skewed toward male cells, male animals and male trial participants, often on the assumption that sex differences were noise to be averaged away — even though women mount generally stronger immune responses to infection and vaccination, at the price of higher rates of autoimmune disease. Funding agencies now require investigators to treat sex as a biological variable, and clinical analyses have hinted at divergent outcomes between men and women receiving checkpoint blockade. What has been missing is a mechanistic, cell-by-cell account of where those differences reside. The authors argue that cancer research remains largely focused on overall populations, overlooking the impact of sex, and their atlas is an attempt to close that gap across the disease spectrum rather than within a single tumor class.

Technically, the study rests on single-cell RNA sequencing, a method that dismantles a tissue into its constituent cells and reads the transcriptome of each one individually. Researchers dissociate tumor samples, capture thousands to hundreds of thousands of cells, barcode their messenger RNA, and sequence it in parallel; computational pipelines then cluster the resulting profiles into discrete cell types and subtypes, allowing investigators to census a tumor’s immune and stromal inhabitants with a granularity that bulk RNA sequencing — which averages signals across entire tissue fragments — cannot approach. Applied as a pan-cancer atlas, the method allowed the team to compile samples from fourteen human cancer types and systematically profile the tumor microenvironment in males and females side by side. Rather than asking how one cancer differs by sex, the design asks which features recur across the disease spectrum — a strategy that helps separate cancer-type idiosyncrasies from sex-linked programs that operate broadly. The census pointed, consistently, toward a specific triad of cell populations expanded in female tumors: myofibroblastic cancer-associated fibroblasts, SPP1-positive macrophages, and exhausted CD4-positive T cells, all of which are implicated in constructing an immune-resistant ecosystem.

Each member of that triad carries a specific mechanistic charge. Myofibroblastic cancer-associated fibroblasts, or myCAFs, are stromal cells that adopt a contractile, matrix-producing state; they lace tumors with dense extracellular matrix, stiffen the tissue, and can physically wall off T cells while secreting factors that suppress antitumor immunity. Macrophages, the tissue-resident scavengers of the immune system, are famously plastic — the same cell can promote tissue repair or stoke inflammation depending on the signals it receives, and within tumors, subsets of these cells are notorious for being co-opted by the malignancy itself. SPP1-positive macrophages, defined by high expression of the gene encoding secreted phosphoprotein 1 — better known as osteopontin, a secreted matricellular protein implicated in metastasis, cell survival signaling and immunosuppression — are increasingly recognized as markers and drivers of immunologically “cold” tumors that resist checkpoint blockade. Exhausted CD4-positive T cells, meanwhile, represent a dysfunctional state of the immune system’s helper arm: chronically stimulated, they lose productive effector function and express inhibitory receptors, contributing to a microenvironment with little for PD-1 blockade to reinvigorate. That all three populations were elevated in female tumors across fourteen cancer types suggests a convergent architecture of immune suppression assembled, at least in part, along sex-specific lines.

The atlas went beyond simply counting cells. Because single-cell data allow researchers to infer communication between cell types from co-expressed ligand-receptor pairs, the team mapped the crosstalk radiating from SPP1-positive macrophages and found it to be conspicuously sex-biased. In female tumors, the SPP1 macrophage module emerges as a signaling hub, wired into fibroblast and T-cell programs in ways that could plausibly coordinate the very triad the census had revealed: macrophages that recruit or polarize fibroblasts, fibroblasts that exclude or exhaust T cells, and T cells that in turn feed back into the suppressive loop. If the wiring diagram of a tumor differs by sex, the authors reasoned, so might its response to a therapy that depends on precisely that circuitry. The study’s title — “Pan-cancer single-cell atlas reveals sex-biased SPP1+ macrophage-associated crosstalk and its implications for immunotherapy” — compresses this logic into a single line: the sex difference is not merely a matter of cell counts, but a difference in the communication architecture of the tumor, with SPP1-positive macrophages sitting at a junction that carries unusually heavy traffic in females.

That hypothesis moved directly into the laboratory. The researchers compared anti-PD-1 immunotherapy responses between male and female mice and observed a clear divergence: the therapy was significantly more effective in males than in females, mirroring the direction of the human atlas findings. They then deployed a genetic model in which Spp1, the mouse equivalent of SPP1, was deleted specifically in macrophages. The intervention had a striking effect: macrophage-specific Spp1 knockout sensitized female mice to anti-PD-1 therapy, narrowing a therapeutic gap that had separated the sexes. The result matters because it crosses the line from correlation to causation. PD-1 blockade works by releasing a molecular brake on T cells, but it can only succeed if the surrounding microenvironment permits reinvigorated T cells to operate; the experiment indicates that in females, SPP1-producing macrophages are not passive bystanders but active enforcers of therapy resistance. Deleting their signature gene did not merely shift biomarkers — it changed therapeutic outcomes, the strongest form of evidence that a cell state is functionally responsible for a phenotype, at least within the bounds of a mouse model.

The work was grounded in real patients as well as large-scale datasets: tumor samples were obtained, with written informed consent, from colorectal and gastric cancer patients operated on at West China Hospital, and all animal procedures were approved by the institution’s ethics committee. For the authors, the implications reach beyond any single cell type. Their conclusion states plainly that sex shapes tumor microenvironment composition and drives divergent therapeutic responses, establishing sex as a critical variable in personalized cancer care and supporting sex-informed strategies to improve immunotherapeutic outcomes. In practical terms, that could mean stratifying patients by sex in immunotherapy trials, interpreting microenvironment biomarkers differently for men and women, or designing combination regimens — for instance, pairing checkpoint blockade with agents that target the SPP1/osteopontin axis or myCAF-driven suppression in female patients whose tumors are enriched for these features. It could also shape how existing drugs are dosed and sequenced, and how immune-related side effects are interpreted, since the immune system’s behavior in one sex cannot simply be mapped onto the other. Osteopontin-targeting approaches are already being explored in oncology, though none has yet reached routine clinical use.

Important caveats temper the excitement. The therapy experiments were performed in mice, and rodent immunology notoriously diverges from the human version; the atlas, however vast, is observational, and in humans it demonstrates association rather than causation. The study also does not settle the deeper question of what drives the sex bias — whether sex hormones such as estrogen and testosterone tune macrophage and fibroblast states, whether X- and Y-linked genes act directly within tumor-infiltrating cells, or whether environmental and behavioral factors layered on top of biology contribute. SPP1 itself is a multitasking protein with established roles in bone maintenance and wound healing, so systemic inhibition could carry liabilities that macrophage-specific genetic deletion in mice does not reveal. And the article is being released as an early, peer-reviewed accepted version that will later be replaced by the final record — a routine publishing practice that nonetheless invites the usual caution before findings harden into clinical doctrine. None of this undermines the central observation, but it frames the discovery as a starting point rather than a destination.

Even with those caveats, the study lands as a pointed challenge to a field that has long treated sex as a nuisance variable. An atlas of 1,662 tumors says otherwise: sex is written into the cellular architecture of cancer, into the crosstalk between macrophages, fibroblasts and T cells, and into the outcomes of the most celebrated therapies of the past decade. The next steps — testing whether SPP1-positive macrophages predict response in human trials, and whether silencing them rescues patients who currently do not benefit — will determine whether the finding becomes a footnote or a fixture of oncology practice. For now, the message is blunt: two patients with the same cancer type, sitting in the same clinic, may carry tumors built on different blueprints. Modern immunotherapy was engineered to release the brakes on T cells, but it will only work for everyone once it accounts for the fact that the road itself looks different depending on the driver’s sex.