Serum sHLA-G stays flat in most endometrial cancer cases
A retrospective case-control study from Kiang Wu Hospital in Macau found that serum soluble HLA-G did not separate women with endometrial cancer from age-matched controls with benign endometrium. The finding matters because sHLA-G has been studied as a candidate immune-evasion signal in gynecologic malignancy, where tumor cells may escape immune recognition by manipulating inhibitory pathways rather than relying only on conventional anatomic spread. Here, however, the blood marker did not rise in a way that would make it useful as a broad diagnostic signal for endometrial cancer. In practical terms, the biomarker failed at the most basic comparison: malignant versus benign tissue context.
The cohort included 55 patients with endometrial cancer and 56 control subjects. Serum sHLA-G was quantified by ELISA, while tissue expression was assessed by immunohistochemistry. The study also examined clinicopathological variables after surgery, including histologic subtype, differentiation grade, myometrial invasion, lymphovascular space invasion, and lymph node involvement. Across the full cohort, no statistically significant difference emerged between cancer cases and controls, and serum levels did not track with several standard markers of local tumor aggressiveness. A blood test that does not move with these features has limited immediate use as a stand-alone biomarker.
Aggressive histology produced the only positive signal
One small subgroup did stand out. Patients with aggressive histological subtypes had higher serum sHLA-G than those with non-aggressive histology, with statistical significance in that comparison. The aggressive subgroup was very small, with only six patients, so the signal should be treated as hypothesis-generating rather than practice-changing. Small subgroup findings can point toward biology, but they can also reflect instability driven by sample size, case mix, or chance. Here, the signal is intriguing because it fits a plausible immunologic model, yet the study design does not allow the marker to be called a validated predictor of behavior.
The biological rationale is straightforward. HLA-G is an immunomodulatory molecule associated with suppression of immune responses, and soluble forms circulating in serum could reflect a tumor microenvironment that is more adept at evading immune attack. In aggressive tumors, that immune escape may be more pronounced, which would explain why serum levels might drift upward in a subset of biologically more aggressive cancers. Still, plausibility is not proof, and the study does not establish whether the serum signal is driven by tumor secretion, host response, or both.
Standard pathology outperformed the biomarker
Among 53 patients with endometrioid carcinoma, serum sHLA-G did not differ meaningfully across grades G1, G2, and G3. The same lack of association held in the subgroup of 36 women who underwent complete surgical staging. Serum levels did not correlate with depth of myometrial invasion, and they were comparable whether or not lymphovascular space invasion was present. The marker also failed to distinguish patients with lymph node metastasis from those without nodal disease.
Those negative findings are clinically important because myometrial invasion, LVSI, and nodal spread remain central to surgical staging and postoperative risk assessment. A circulating biomarker earns its place only if it adds information beyond histology and anatomy. In this cohort, sHLA-G did not do that. The data suggest that routine pathology still carries the prognostic load, while serum sHLA-G remains too blunt to replace standard risk stratification.
Tissue expression and blood levels did not align
The study also found no significant association between serum sHLA-G concentration and tissue HLA-G immunohistochemical status. That lack of concordance weakens the argument for using a single serum measurement as a surrogate for tumor expression. Tumor biology in endometrial cancer is heterogeneous, and immune-related markers may vary across the primary lesion, invasive front, and metastatic deposits. A blood sample, by contrast, compresses that complexity into one circulating snapshot.
When blood and tissue markers fail to align, the most cautious interpretation is that serum reflects only part of the underlying biology, or possibly none of the relevant biology at all. The disconnect also suggests that immunohistochemical positivity alone does not guarantee a measurable circulating signal. For a biomarker to be clinically useful, it needs both biological relevance and reproducibility across compartments. This study did not show either in a way strong enough to support immediate clinical translation.
What the study can and cannot support
The most defensible conclusion is narrow. Serum sHLA-G was not elevated in endometrial cancer overall, did not correlate with tumor grade or key surgical pathology variables in endometrioid carcinoma, and did not match tissue HLA-G status. The only positive association appeared in an exploratory analysis of aggressive histologic subtypes, and even that result rests on a very small sample. No one should treat the finding as evidence that sHLA-G is already a diagnostic or prognostic biomarker.
The study design also places guardrails around interpretation. It was retrospective, case-control, and single-center, with all the usual limitations of such work: potential selection effects, a modest sample, and subgroup analyses that can easily overfit sparse data. The ROC analysis performed for the aggressive subgroup was explicitly exploratory and lacks external validation. Without replication in a larger independent cohort, the assay cannot be used to guide treatment decisions or surveillance.
For clinicians, the message is conservative but useful. Endometrial cancer risk stratification still depends on histology, depth of invasion, LVSI, and nodal status, not on circulating sHLA-G. For researchers, the signal in aggressive histology may justify further study of immune evasion pathways, but only with a larger cohort and stronger validation strategy. Until then, the biomarker remains a candidate, not a tool.