Tumor vascularity--a novel prognostic factor in advanced cervical carcinoma.

OBJECTIVE

In the search for the optimal treatment of advanced cervical cancer, the identification of valid prognostic factors obtainable without histopathologic investigation of the entire tumor and the locoregional lymph nodes is of paramount interest. Tumor microvessel density has recently been demonstrated to correlate strongly with disease aggressiveness in breast cancer and other malignancies.

METHODS

We established a computerized image analysis system to quantify tumor microvascularity by using the closest-individual method, which determines the distribution of distances from random points within the tumor to the closest microvessel (DTCMV). Tumor microvascularity was assessed in paraffin sections of two cylindrical 2 x 20-mm core biopsies obtained transvaginally from the 12 and 6 o'clock positions of each tumor and then immunohistochemically stained for Factor VIII-related antigen. The oncologic relevance of tumor vascularity is studied in an open prospective trial.

RESULTS

Tumor vascularity was quantified in 42 patients with cervical cancers > 3 cm in largest diameter, FIGO stages Ib-IVa. This new parameter representing pathophysiological tumor-host interactions was independent of various other patient and tumor characteristics, including age, FIGO stage, tumor size, differentiation, lymph node metastases and lymphatic space involvement. Thirty-nine patients were treated with curative intent either by primary surgery (n = 22) or radiation (n = 17). After a median observation time of 18 months (range 4-41 months), the patients with higher tumor vascularity (mean DTCMV < 83 microns) had significantly shorter disease-free (P = 0.025) and overall (P = 0.032) survival probabilities than patients with lower tumor vascularity (mean DTCMV > or = 83 microns). Cox regression analysis identified tumor vascularity as the strongest independent prognostic factor in this group of patients.

CONCLUSIONS

The assessment of tumor microvascularity by computerized image analysis of defined tumor biopsies could become a novel means of predicting tumor aggressiveness in non-early cervical cancer.