Nuclear Shape Analysis for Assessment of Prognosis in Renal Cell Carcinoma

Abstract

Clinically localized renal cell carcinoma is cured by radical nephrectomy in 47% [stage T3a (II)] to 65% [stage T1, T2 (I)] of the patients. Local recurrence and metastatic disease probably result from undetectable microscopic metastases present at operation. Chemotherapy and immunotherapy may improve cure rates if administered adjuvantly. The outcome of individual patients who share surgical stage cannot be predicted reliably by tumor histology, pathological and/or nuclear deoxyribonucleic acid analysis. Two groups of 10 patients with clinically localized renal cell carcinoma were similar by sex distribution (5 men and 5 women), surgical stage (stages T1 in 1, T2 in 6 and T3a in 3 patients) and age (54.3 +/- 15.2 standard deviation versus 55.8 +/- 8.7 years). Group 1 had no recurrences with a minimum followup of 5 years and a mean followup of 10 years. Group 2 died of metastatic renal cell carcinoma after a mean of 5 years. All neoplastic areas of each paraffin-embedded operative specimen were randomly sampled and the nuclear perimeter of 150 cancerous cells was digitized. There were 25 shape descriptors calculated for each nucleus. All shape descriptors for each patient were described by 19 statistical tests. Nuclear perimeter and area as well as mean nuclear roundness factor failed to separate the 2 groups. Range median quartiles of ellipticities by Fourier analysis, coefficients of variation of chain code minimums and relative means of largest 10 convexity values produced greatest separation (Mann-Whitney-Wilcoxon test p less than 0.001, and variance normalized difference 3.21, 3.29 and 2.83, respectively). These descriptors normalized and summed provided near perfect separation (Mann-Whitney-Wilcoxon test p less than 0.001 and variance normalized difference 3.59). We developed a quantitative nuclear morphometric analysis system that permitted the correct assignment of outcome in 19 of 20 patients. Accurate prediction of prognosis in patients with clinically localized renal cell carcinoma by nuclear shape analysis may allow for selection of patients for adjuvant therapy who have clinically undetectable metastatic disease.

Prognostic Significance of Granular Cell Content in Renal Cell Carcinoma

“…This method in their opinion may prove superior to stag ing because it reflects the biological potential of the neo plasm even before its clinical expression (staging). Re cently, Murphy et al [11] proposed nuclear shape analysis for the prognosis of RCC.…”

Decreased expression of α-catenin is associated with poor prognosis of patients with localized renal cell carcinoma

“…Thus, immunohistochemical analysis using anti-␣-catenin antibodies may be used to identify high-risk patients. As 45-65% of patients with localized RCC (pT1-3, N0, M0) will develop a local recurrence or metastases after radical nephrectomy (Patel and Livengood, 1977;Murphy et al, 1990) and prognosis of these patients is very poor with a mortality of approximately 74% at 1 year after tumor recurrence (Patel and Livengood, 1977), accurate identification of high-risk patients is indeed relevant.…”

“…However, within the group of low-stage tumors (pT1-3, N0, M0), a significant number of patients show recurrence of disease after radical nephrectomy (McNichols et al, 1981). Several investigators have studied the value of possible prognostic indicators, e.g., grade of nuclear pleomorphism (Murphy et al, 1990), morphometric analysis (van der Poel et al, 1993), DNA flow cytometry (Ljungberg et al, 1986;Oosterwijk et al, 1988), proliferation rate (Chauhan et al, 1991;de Riese et al, 1993;Papadopoulos et al, 1994;Flint et al, 1995), p53 over-expression (Bot et al, 1994;Oda et al, 1995), expression of renal antigens (Oosterwijk et al, 1988) and specific genetic alterations (Nanus et al, 1990;Foster et al, 1994;Kovacs, 1994). Despite intensive research, no clear marker has been identified that can distinguish patients at risk for tumor recurrence within the pT1-3 group.…”

Morphometric index of adult renal cell carcinoma

“…They used the Euclidean distance calculation (square root of the sum of the squared differences in values for each feature) to show the existence of two subgroups of RCC differing in their degree of similarity to normal. Murphy et al [34] used the Manhattan distance measure to calculate the degree of separation between groups of RCC patients with different survival times.…”