The study, published in BMC Genomics, took advantage of the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be, known as nuMoM2b, a multi-center prospective cohort that collected biospecimens in early pregnancy, genome-wide genotype data, and detailed clinical outcomes across multiple ancestry groups. Led by researchers at Indiana University and collaborators across the United States, the team performed genome-wide association studies of first-trimester CRP levels using maternal genotypes from 4,326 participants and offspring genotypes from 2,140 children. The design allowed them to ask two distinct questions at once: which parts of the genome influence CRP levels during early pregnancy, and whether the inherited propensity toward higher or lower CRP translates into a measurable shift in gestational diabetes risk.

The first answer came cleanly. In the European-ancestry maternal sub-cohort, the researchers identified three genome-wide significant loci associated with early-pregnancy CRP levels: the CRP gene itself, LEPR, which encodes the leptin receptor, and HNF1A, a transcription factor with a well-documented role in regulating hepatic CRP production. These findings align closely with what large GWAS in non-pregnant populations have reported, indicating that the fundamental genetic architecture of CRP regulation persists into pregnancy. That consistency matters, because it suggests the pregnancy state does not wholesale rewrite the genetic control of this inflammatory marker, even though pregnancy itself dramatically reshapes inflammatory physiology.

When the analysis was widened to the full multi-ancestry maternal cohort, two additional loci emerged: ENSG00000257703 and APOC1. The appearance of ancestry-informative signals underscores a recurring theme in human genomics: cohorts that concentrate on a single ancestry group miss variants that are common or consequential elsewhere. APOC1, involved in lipid metabolism, sits in a genomic region long known to influence CRP, and its detection here hints at connections between inflammatory and metabolic pathways that pregnancy may bring into sharper focus. By contrast, the offspring GWAS of CRP levels produced no genome-wide significant associations at all, a null result the authors interpret in light of the smaller sample size and the distinct biology of fetal and neonatal CRP regulation.

The more provocative part of the study concerns gestational diabetes. Using linkage disequilibrium score regression, a technique that estimates the genetic correlation between traits from summary statistics alone, the team found no significant genetic correlation between CRP and gestational diabetes when using the nuMoM2b CRP GWAS as the input. That absence of correlation would seem to close the door on the idea that genetically influenced CRP levels meaningfully shape GDM risk. But the picture grew more complicated: when the researchers substituted an external, much larger population-based CRP GWAS, a significant genetic correlation with gestational diabetes did appear.

That discrepancy is not a contradiction so much as a lesson in how context shapes genomic inference. Genetic correlations estimated from summary statistics can vary depending on the population studied, the environment in which the phenotype was measured, and the specific covariates and ascertainment of each cohort. A CRP GWAS measured in early pregnancy among nulliparous women captures a different biological moment than one measured in the general adult population, where inflammation reflects age, adiposity, infection history, and chronic disease. The authors suggest that the genetic overlap between CRP and gestational diabetes may therefore vary across study contexts, a caution that extends well beyond this particular pair of traits.

To probe the relationship more directly, the team constructed polygenic risk scores for CRP, aggregating the small effects of thousands of variants into a single inherited score per individual, and tested whether those scores predicted gestational diabetes in the nuMoM2b cohort. They did not, regardless of whether the scores were derived from the pregnancy-specific GWAS or from the external population-based one. In other words, even where a statistical genetic correlation could be detected at the level of populations, the aggregate inherited influence on CRP was not a useful predictor of who would develop gestational diabetes within this cohort. The polygenic signal, the study concludes, is limited.

For clinicians and researchers, the result is a sobering check on a tempting hypothesis. Elevated CRP in early pregnancy is associated with later gestational diabetes in observational studies, and it has been natural to wonder whether inflammation is part of the causal chain. This work indicates that the genetic component of CRP variation, at least as it can currently be measured, contributes little to that association. If CRP and gestational diabetes are connected, the link is more likely mediated by environmental factors, adiposity, insulin resistance, or the hormonal shifts of pregnancy than by inherited differences in inflammatory set-point. Genetic risk prediction for gestational diabetes will need to look elsewhere, most plausibly toward the substantial polygenic architecture of glycemic traits themselves.

The study also carries a broader methodological message. Pregnancy-specific genomic resources remain scarce relative to their clinical importance, and this analysis demonstrates both their value and their limits. The nuMoM2b cohort, with thousands of well-phenotyped pregnancies and paired maternal-offspring genotypes, was large enough to replicate known CRP loci and to detect new ones in a multi-ancestry framework, yet the authors are explicit that larger pregnancy cohorts are needed to fully untangle the relationship between inflammation and metabolic complications of pregnancy. Gestational diabetes affects millions of pregnancies worldwide each year, and its long-term sequelae, including elevated lifetime risk of type 2 diabetes in mothers and altered metabolic programming in offspring, make it a priority target for precision approaches.

What emerges from this work is a carefully bounded conclusion rather than a dramatic one. The genetic architecture of CRP in early pregnancy mirrors what has been mapped in non-pregnant populations, with CRP, LEPR, and HNF1A as anchor points and APOC1 and one additional locus joining the map in a multi-ancestry analysis. Offspring CRP genetics yielded no significant signals. Genetic correlation with gestational diabetes appears only under some analytic conditions and not others, and polygenic risk scores for CRP fail to predict the disease. In an era when inflammatory biomarkers are frequently proposed as early warning signs of pregnancy complications, the study offers a useful corrective: a biomarker can travel with a disease without its genes being responsible for it, and disentangling the two requires cohorts, ancestries, and analytic frameworks designed specifically for the biology of pregnancy.

Subject of Research: Genome-wide association study of C-reactive protein levels in pregnancy and its polygenic relationship with gestational diabetes mellitus

Article Title: Maternal and offspring genome-wide association study of C-reactive protein reveals limited polygenic association with gestational diabetes mellitus

Article References: Zhang, Y., Moore, A., Ryckman, K. K., Yan, Q., Guerrero, R. F., Li, M., Silver, R. M., Luo, J., Yee, L. M., Reddy, U. M., Feghali, M. N., Chung, J., Haas, D. M., Kua, K. L., & Liu, N. (2026). Maternal and offspring genome-wide association study of C-reactive protein reveals limited polygenic association with gestational diabetes mellitus. BMC Genomics, 27(1), Article 782. https://doi.org/10.1186/s12864-026-12878-6

Image Credits: AI Generated

DOI: 10.1186/s12864-026-12878-6

Keywords: C-reactive protein, gestational diabetes, GWAS, polygenic risk score, nuMoM2b, pregnancy, inflammation, genetic correlation, LEPR, HNF1A, APOC1, maternal-fetal health

The research builds on the group’s earlier work showing that myeloid-derived suppressor cells, or MDSCs, accumulate in abundance within human esophageal squamous cell carcinoma, the dominant histological subtype of esophageal cancer worldwide, and that their presence correlates with worse patient outcomes. What remained unclear was whether MDSCs contribute to tumor propagation through mechanisms that go beyond their canonical role of suppressing T cells. To answer that question, the investigators turned to mouse models of esophageal cancer that included both immunocompetent animals and mice lacking T cells altogether. If MDSCs promoted tumor growth only by disarming T cells, their influence should have vanished in the T cell-deficient setting. Instead, the tumors accelerated in both contexts, a result that signaled the existence of an intrinsic, T cell-independent tumor-promoting program driven by these myeloid cells.

Tracking down the molecular machinery behind that program led the team to an unexpected transcription factor. When the researchers depleted MDSCs in tumor-bearing mice using an anti-Gr1 antibody, or when they neutralized the inflammatory cytokine interleukin-1 beta, expression of HNF1A inside the tumor cells dropped. HNF1A, a transcription factor best known for its roles in liver and pancreatic development and metabolism, has more recently been implicated in tumor biology, but its involvement in MDSC-driven esophageal cancer progression had not been established. The new data place it at the center of the circuit: MDSCs, acting through IL-1 beta, appear to switch on HNF1A within the carcinoma cells themselves.

From HNF1A, the signal flows onward to a chemokine. The study showed that HNF1A sustains the activity of the CXCL1 promoter, and that disrupting the upstream inputs, whether by depleting MDSCs or blocking IL-1 beta, blunted CXCL1 promoter activity and reduced CXCL1 output. CXCL1 is a chemokine of the CXC family that signals through the receptor CXCR2 and is well recognized as a mediator of neutrophil recruitment and inflammatory signaling in tumors. In this context, however, its downstream consequences are not primarily inflammatory in the classical sense. Rather, the researchers found that the HNF1A-CXCL1 axis feeds directly into the core stemness machinery of the cancer cells, downregulating or, when the axis is active, sustaining the expression of the transcription factors Oct4, Nanog and Sox9, the canonical guardians of stem-like identity in embryonic and cancer stem cells alike.

The functional consequences of that molecular cascade were tested with rigor. When the team deleted HNF1A specifically in tumor cells, the MDSC-driven expansion of the cancer stem cell population collapsed, and, critically, the ability of the tumors to initiate new growth was abolished. Tumor initiation is the hallmark functional readout of cancer stem cell activity: only cells with genuine self-renewal capacity can seed a new tumor from a limited inoculum. The fact that HNF1A deletion eliminated this capacity demonstrates that the transcription factor is not merely a correlate of stemness but a required licensing factor in this pathway. Conversely, when the researchers supplied exogenous IL-1 beta, the cancer stem cell frequencies rebounded, confirming that the myeloid-cell-derived cytokine sits upstream of the entire axis and is sufficient to re-ignite the stem-like program.

These results reframe the relationship between inflammation and cancer stemness in esophageal cancer. Rather than acting as passive bystanders that merely modulate the immune microenvironment, MDSCs emerge as active instructors of tumor cell identity, delivering an IL-1 beta signal that is transcribed into a heritable, self-reinforcing stem-like state through HNF1A and CXCL1. Because the pathway was operative in T cell-deficient mice, the authors conclude that MDSC-derived IL-1 beta licenses the HNF1A-CXCL1 axis and sustains cancer stem cell properties independently of adaptive immunity. That independence matters clinically. Modern oncology has invested heavily in T cell-directed strategies, from immune checkpoint inhibitors to engineered cell therapies, and esophageal cancer has been among the tumor types to benefit. But a tumor-promoting program that runs beneath the T cell layer provides a reservoir of malignant potential that such therapies would not touch, and it may help explain why responses in esophageal squamous cell carcinoma remain incomplete for many patients.

The therapeutic implication drawn by the authors is that targeting this pathway could complement, rather than replace, T cell-focused treatments. Several points of intervention suggest themselves from the data. MDSC depletion, as achieved with anti-Gr1 in the preclinical setting, removes the source of the signal. IL-1 beta neutralization interrupts the messenger, and IL-1-blocking agents already exist in the clinical arsenal for inflammatory diseases, offering a plausible route to translation. Downstream, the CXCL1-CXCR2 axis is a recognized drug target, with CXCR2 antagonists under investigation in multiple cancers. Each of these strategies was supported, directly or indirectly, by the experimental results: depleting MDSCs or neutralizing IL-1 beta reduced HNF1A expression, dampened CXCL1 promoter activity and lowered stemness factor levels, while restoring IL-1 beta reinstated cancer stem cell frequencies.

The study also carries prognostic weight. The graphical summary accompanying the article emphasizes that the HNF1A-driven CXCL1 program not only licenses stem-like properties in esophageal squamous cell carcinoma but also predicts poor prognosis, consistent with the team’s earlier finding that MDSC abundance in human tumors correlates with adverse outcomes. Cancer stem cells are widely associated with resistance to chemotherapy and radiotherapy, metastatic dissemination and relapse after apparently curative treatment, so a microenvironmental signal that expands this compartment provides a mechanistic bridge between inflammatory infiltration and clinical aggressiveness. For patients with esophageal squamous cell carcinoma, a disease with persistently poor survival statistics in many regions, that bridge may represent one of the more actionable links identified to date.

Methodologically, the work combined genetically defined mouse models, cell-specific deletion of HNF1A, pharmacologic MDSC depletion, cytokine neutralization and supplementation, and molecular readouts of promoter activity and stemness factor expression, an integrated design that allowed the authors to move from correlation to causal mechanism. The study was supported by the National Natural Science Foundation of China, and the MEC25 cell line used in the experiments was provided by the laboratory of Professor Li Fu at Shenzhen University Medical School. The corresponding author, Yi Zhang, holds appointments across the Biotherapy Center and Cancer Center of the First Affiliated Hospital of Zhengzhou University, the State Key Laboratory of Metabolic Dysregulation and Prevention and Treatment of Esophageal Cancer, and related Zhengzhou University institutions, reflecting the translational infrastructure behind the project. The article was received in January 2026, accepted in September 2026 and published on 20 September 2026 under a Creative Commons Attribution license.

As with any preclinical study, the path from mouse models to patient benefit will require validation in human tumor specimens and, ultimately, clinical testing of pathway-targeted interventions. Nevertheless, the conceptual advance is clear and consequential: the immune microenvironment does not merely decide whether the immune system sees a tumor, it can also decide what the tumor is. In esophageal squamous cell carcinoma, myeloid-derived suppressor cells appear to whisper a developmental command into the cancer cells, activating HNF1A, broadcasting CXCL1 and preserving a self-renewing core that survives whatever the immune system or the oncologist throws at it. Silencing that command, whether by removing the myeloid messengers, intercepting their IL-1 beta message or blocking the CXCL1 relay downstream, now stands as a defined and testable strategy to strip esophageal cancer of its stem-like resilience and to make T cell-directed therapies work against a smaller, more vulnerable target.

Subject of Research: How myeloid-derived suppressor cells confer cancer stem cell properties to esophageal squamous cell carcinoma via the HNF1A/CXCL1 signaling axis

Article Title: Myeloid-derived suppressor cells confer stemness to esophageal cancer cells through the HNF1A/CXCL1 signaling axis

Article References: Qin, G., Ma, P., Liu, S., Chen, T., Guo, K., Zhao, Q., Wu, P., Chen, X., & Zhang, Y. (2026). Myeloid-derived suppressor cells confer stemness to esophageal cancer cells through the HNF1A/CXCL1 signaling axis. Cancer Immunology, Immunotherapy. https://doi.org/10.1007/s00262-026-04577-8

Image Credits: AI Generated

DOI: 10.1007/s00262-026-04577-8

Keywords: myeloid-derived suppressor cells, esophageal squamous cell carcinoma, HNF1A, CXCL1, cancer stem cells, interleukin-1 beta, tumor microenvironment, stemness, Oct4, Nanog, Sox9, cancer immunology