In our study, patients with the micropapillary urothelial variant showed potential benefits from the COBRA score, which can refine clinical prognosis predictions. This stands in contrast to other urothelial variants where the correspondence was not as precise. Such insights can guide the tailoring of postoperative therapy and surveillance.

Nomograms can incorporate and simultaneously account for the effects of multiple factors, improving predictive accuracy and, subsequently, individual care and counseling [13]. A model specifically for bladder cancer with UV may benefit evidence-based care. Each UV represents a unique neoplasm with different underlying genetics and biology, different natural history, and different oncological outcomes [14].

The COBRA score was developed by Welty et al. to assess the postoperative prognosis after RC for patients with urothelial carcinoma of the bladder [6] using clinical and pathological characteristics. In our study, the micropapillary variant showed the best correlation with the COBRA and modified COBRA scores [15]. The AUC of COBRA score was 68% for OS, 63% for CCS 2 years, and 62% for CCS in 5 years. The original score description similarly presents an AUC of 68% for predicting OS [6]. However, the score was designed without information on CSS, which we assessed in our study.

The original score was previously modified to incorporate lymphovascular invasion (LVI) to improve score accuracy, based on 789 patients with a median follow-up of 32 months [8]. We also observed better accuracy with the modified COBRA score. In another study, the COBRA score model was validated with 914 patients and a median follow-up of 29 months. The C-index of the COBRA score and COBRA score stratified were 0.69 and 0.71 to CSS in 2 years, similar to our results [9].

Several prognostic models have been published to predict oncological outcomes after RC [15,16,17]. Shariat and Karakiewicz developed a nomogram that included age, sex, tumor grade, lymphovascular invasion, carcinoma in situ, neoadjuvant chemotherapy, adjuvant chemotherapy, and adjuvant external beam radiotherapy. However, multiple variables make it more difficult to use in clinical practice. The COBRA score modified with LVI is effective, practical and facilitates the more widespread use of the nomogram [8].

The modified COBRA score and stratified COBRA scores were all applied to the UV patients in our population. However, only the micropapillary variant showed good correlation and accuracy using these prognostic nomograms. The pathologist Amin at MD Anderson described micropapillary urothelial bladder cancer in 1994 [18], and lymph node involvement in this histological subtype seems more prevalent than in other variants, presenting higher lymph node density than other UV [19]. This important role of lymph node involvement could explain the good prognostic correlation of micropapillary UV with the COBRA score.

A total of 19 patients presented non-muscle invasive bladder tumors in the anatomopathological analysis of specimens of radical cystectomy (10 patients with squamous cell carcinoma, five patients with micropapillary, two patients with glandular carcinoma, one lymphoepithelioma like and one undifferentiated carcinoma). The expressive number of patients with non-muscle invasive micropapillary variant (26.31%) could influence the survival analysis and the correspondence with the COBRA score.

There are limitations of our study inherent to its retrospective study design. All reported cases presented at least 10% of the histological variant in the analyzed samples. However, the imprecision in the percentage of tissue involved by the variant may add another degree of uncertainty. In future, it appears essential to refine the representation of UV in the samples to gain clearer insight into its clinical prognostic implications. In addition, the small sample size and limited follow-up make it difficult to generalize these results to other populations.

We must consider that the retrospective character on a current theme that has undergone significant conceptual changes over time is also a bias to be considered since the cases was analyzed from 2009 to 2022.

Lastly, the low number of patients who completed neoadjuvant chemotherapy, which can be attributed to a paradigm shift within the public institution and the logistical challenges faced by patients in accessing the oncological reference center, and this may have influenced survival analysis.

The COBRA score must be prospectively validated in patients with UV, and its high predictive accuracy must be confirmed before it can be incorporated into the clinical routine.