Ovarian Cancer T-Cell-Macrophage States Track Chemo Response
Key Takeaways
- In a very small treatment-naive ovarian cancer cohort of fewer than 20 patients, predicted T-cell-macrophage doublets were dominated by CD8+ interactions and were relatively more enriched in chemotherapy-sensitive than resistant cases.
- Resistant-associated T-cell-macrophage doublets carried T-cell exhaustion and M2 macrophage features, whereas sensitive-associated doublets carried M1 macrophage enrichment with relatively lower exhaustion.
- In a separate spatial transcriptomics cohort, T-cell-macrophage spots were about 5 times more frequent in good responders than in partial responders, and CD8+ T-cell-macrophage spots were seen only in good responders.
- Enriched ligand-receptor patterns in T-cell-macrophage doublets included HLA-TCR and ICAM- or VCAM-integrin families consistent with antigen presentation and immune synapse biology.
In Hameed et al. in the International Journal of Molecular Sciences, a translational analysis paired ovarian cancer single-cell RNA sequencing (scRNA-seq) with a separate spatial transcriptomics cohort. The main dataset included 14 patients overall, including 13 treatment-naive patients who underwent surgery followed by platinum-based chemotherapy; resistance was defined as progression within 6 months after adjuvant therapy, yielding 4 resistant and 9 sensitive cases. Investigators used an annotated singlet reference of about 220,000 cells and a doublet-retained dataset of about 520,000 cells after preprocessing, and applied uniform linear model network (ULMnet) to infer physical cell-cell interaction networks. Downstream analyses centered on predicted CD4-positive and CD8-positive T-cell-macrophage doublets, while the spatial cohort was a separate interval-debulking cohort of 8 ovarian cancer patients sampled after 3–5 cycles of neoadjuvant chemotherapy and grouped as good, partial, or poor responders by chemotherapy response score, with T-cell-macrophage spots defined by more than 10% T cells and more than 10% macrophages.
The analysis identified 1323 CD8-positive T-cell-macrophage doublets and 384 CD4-positive T-cell-macrophage doublets. CD8-positive doublets predominated overall and were relatively more enriched in sensitive cases, whereas resistant cases had a higher relative share of CD4-positive doublets. Resistant-associated doublets carried T-cell exhaustion and M2 macrophage features, while sensitive-associated doublets carried M1 macrophage enrichment with relatively lower exhaustion. Stemness was higher in CD4-positive doublets, whereas cytotoxicity was higher in CD8-positive doublets.
In the spatial cohort, T-cell-macrophage spots were about 5 times more frequent in good responders than in partial responders and were negligible in poor responders. CD8-positive T-cell-macrophage spots appeared only in good responders. Ligand-receptor patterns were consistent with antigen-presentation and immune-synapse programs, including human leukocyte antigen-T-cell receptor and ICAM- or VCAM-integrin families. Doublet-to-spot enrichment correlations were r = 0.33 for sensitive or good-response patterns and r = 0.27 for resistant or bad-response patterns, although this provided only partial, indirect cross-cohort support because the spatial groups reflected chemotherapy response scores after 3–5 cycles of neoadjuvant therapy rather than post-adjuvant progression within 6 months.
The authors noted that some predicted doublets could reflect random co-encapsulation rather than true physical contacts. They also cautioned that the primary cohort was small and imbalanced across response groups, that stage and sampling-site differences may have influenced the observed patterns, and that the work remained computational with only partial spatial support and no direct experimental validation. Gene and transcription-factor changes within a doublet also could not be assigned definitively to the T-cell or macrophage component. In that context, these interaction states are best framed as translational, hypothesis-generating biology rather than validated predictive biomarkers.
According to the researchers, ovarian T-cell-macrophage interaction states tracked with chemotherapy response patterns, with more favorable response patterns aligning with CD8- and M1-associated states and resistant patterns aligning with exhaustion and M2-associated states. The spatial cohort provided partial support for the modeled interaction landscape.
Clinician Questions
How was chemotherapy resistance defined in the ovarian cancer scRNA-seq cohort?
In the ovarian cancer single-cell RNA sequencing cohort, resistance was defined as disease progression within 6 months after completion of adjuvant platinum-based therapy, and the remaining cases were classified as sensitive.
Which ovarian cancer samples contributed to the inferred T-cell-macrophage interaction map?
The inferred interaction map drew on ovarian cancer samples from primary ovarian tumor, omental metastasis, primary lymph node, malignant ascites, and peripheral blood. The predicted T-cell-macrophage doublets were predominantly derived from malignant ascites; within the smaller tumor-site fraction, CD8-positive doublets were relatively more common at primary and metastatic tumor sites, whereas CD4-positive doublets were more enriched in lymph nodes.
How were T-cell-macrophage spots defined in the spatial ovarian cancer cohort?
In the spatial transcriptomics cohort, a spot was classified as a T-cell-macrophage spot when spot-level deconvolution showed more than 10% T cells and more than 10% macrophages.