Introduction:

Circulating tumor cells (CTCs) are promising biomarkers for risk stratification in urothelial cancer (UC). Recently CTC morphology (assessed via digital pathology) and CTC heterogeneity have emerged as novel biomarkers that are prognostic for survival in patients with metastatic genitourinary cancers. Here, we investigated whether a “bad cell” CTC signature defined previously in a cohort of metastatic urothelial cancer patients was also associated with progression to metastatic disease in a cohort of localized bladder cancer patients. For comparison, we also evaluated CTC heterogeneity and total CTC count as alternate biomarkers.

Methods:

Peripheral blood samples were collected from N=16 consecutive UC patients at baseline and every 3 months (+/-14 days). Patients included 6 (37%) high-risk non-muscle-invasive bladder cancer (NMIBC) patients who had progressed on BCG and enrolled on a trial of vicineum with durvalumab, as well as 10 (63%) muscle-invasive bladder cancer (MIBC) patients (6 pre- and 4 post-cystectomy). Slides were processed with the Epic CTC platform and stained with pan-CK/CD45/PD-L1/DAPI for CTC detection. Morphologic subtype was assigned based on 11 morphologic features including nuclear solidity, speckling, nucleoli and entropy; cytokeratin speckling and ratio; and cytoplasmic/nuclear circularity, area, and convex area ratio. CTC heterogeneity (Fig 1) was measured with the Shannon Index (SI), with optimized cutoff SI >0.6 for high heterogeneity. Kaplan Meier survival analysis was used to evaluate metastatic progression free survival (PFS).

Results:

With median follow-up 8.7 months, 7(44%) patients had metastatic progression (1 NMIBC,6 MIBC). Analysis was limited to the first time point collected as only 8/16(50%) had >=2 time points. CTCs were detected in 80%(4/6) NMIBC and 60%(6/10) MIBC patients. The “bad cell” signature was identified only in the 2 pre-cystectomy MIBC patients who had their cystectomies converted to palliative or aborted due to metastatic disease found intraoperatively(Fig 2A). CTC detection was not associated with PFS(Fig 2B). In contrast, high CTC heterogeneity was associated with 1.3 vs 15.6 months PFS(p<0.0001,Fig 2C). The 8 cases with multiple time points were evaluated for consistency-7/8(88%) cases were consistent for SI at all time points.  All 4 MIBC cases were consistent for SI and "bad cell"; CTC detection was inconsistent in 1 case. For NMIBC, only 1/4 cases was consistent for CTC detection. Notably, 3 NMIBC cases without progression inconsistently displayed the “bad cell” signature.

Conclusion:

Digital pathology and subtype assignment of CTCs is relevant in localized bladder cancer. Here, the “bad cell” signature seemed to identify MIBC patients who were locally advanced at cystectomy, though it was not useful in NMIBC patients. CTC heterogeneity was associated with worse PFS and was consistent across replicate time points.  Detecting CTCs regardless of subtype was not a useful biomarker in this cohort. Further investigation is warranted to validate these findings in a larger cohort of localized bladder cancer patients.

Funding: This research was supported by the Intramural Research Program of the NIH, and the Greenberg Bladder Institute.