Richard A. Greenberg

Evaluation of size in prognosis of oral cancer

Greatest surface diameter of a cancer, together with suspicion of regional node metastasis, forms the basis for prognosis through the clinical TNM staging system for many cancers. In oral cancer, however, surface size sometimes fails to correlate, or sometimes inversely correlates, with tumor aggressiveness. To shed light on the value of measuring size per se, 155 consecutive oral squamous cancers, treated by surgery, radiation, or a combination, were analyzed to find the degree of correlation between greatest surface measurement and pathologic nodal spread and control of cancer. In tumors less than 2 cm, size correlated with very few nodal metastases and with good prognoses; in tumors greater than 2 cm, increasing size did not show a corresponding increase in pathologic node metastasis or significantly worsening outcomes except for a few very large cancers invading adjacent structures. In conclusion, greatest surface diameter of an oral cancer, when greater than 2 cm, is an unreliable predictor of tumor behavior per se. A small pilot study suggests that tumor thickness may be a better predictor. A formal study of this is planned. Cancer 58:158–162, 1986.

Computerized histologic assessment of malignant potential: A method for determining the prognosis of uveal melanomas

The authors evaluated 50 cases of primary melanoma of the choroid and ciliary body. In each case, 100 cells were randomly selected from a single histologic slide, and on each cell computer-assisted measurements were made of 18 nuclear and nucleolar features. Means and standard deviations were calculated for each of these features in each tumor. Thirteen calculated variables (six means and seven standard deviations) were found to correlate significantly with patient mortality following enucleation. Standard deviations of statistically significant nuclear and nucleolar features demonstrated significantly greater correlation with mortality than the means of these features, thus confirming the great value of nuclear pleomorphism for predicting the malignant potential of uveal melanomas. Furthermore, when the standard deviation of the nucleolar circumference, a feature highly correlated with survival (P less than 0.00001), was combined with the measurement of the largest dimension of the tumor, linear discriminant analysis correctly predicted the clinical course of 88 per cent of cases.

Objective assessment of the malignant potential of intraocular melanomas with standard microslides stained with hematoxylin-eosin

A method for extracting prognostic information from the cytologic details of intraocular melanomas was developed. With computer assistance, a technician measures the nucleolar area of 200 tumor cells randomly selected from a standard paraffin-embedded microsection stained with hematoxylin-eosin, a procedure that requires approximately 40 minutes with an apparatus that costs less than +30,000. Standard deviation of nucleolar area (SDNA), computed from these measurements, provides an objective assessment of nucleolar pleomorphism. Previous studies showed that second measurements of the same microsection or of the same tumor at different levels yield SDNA values that closely approximate the original measurements. The application of this method to 540 cases from three independent laboratories yielded a correlation of SDNA with death from tumor that was high (P much less than 0.001) and consistent among cases from all three sources. Thus, SDNA provides a clinically useful cytologic measure of the malignancy of intraocular melanomas and may ultimately prove valuable in the assessment of other types of tumors.

Interval-by-interval cox model analysis of 3680 cases of intraocular melanoma shows a decline in the prognostic value of size and cell type over time after tumor excision

In a selected group of 3680 patients treated by ocular excision for uveal melanoma, 1178 deaths were attributed to this tumor during 20 years following ocular excision. Survival time was divided into intervals containing an equal number of deaths, and coefficients of the Cox statistical model were independently derived for each interval. This analysis revealed a statistically significant decline over time in the prognostic value of largest tumor dimension (LTD) and Callender cell type (CT). Mathematical considerations suggest that the prognostic value of parameters derived from other cancers may also decline with time following excision of the primary tumor.

A clinically useful method for combining gross and microscopic measurements to select high-risk patients after enucleation for ciliochoroidal melanoma

An objective and reproducible method was devised for estimating the malignant potential of ciliochoroidal melanomas after enucleation. This method requires only a singleroutine microslide from which nucleolar area and largest tumor dimension are measured. Previous studies have shown these measurements to be reproducible and highly correlated with mortality. To demonstrate a practical method for clinical application of these measurementsa Cox statistical model was derived from 200 cases supplied by the Armed Forces Institute of Pathology. The resulting modelwhen applied to 340 cases with known outcome from two independent laboratoriesallowed subdivision of patients into groups that suffered a six‐fold difference in mortality. These results suggest that a central registryby applying this method to histologic slides from around the worldcould provide information useful for the clinical management of patients enucleated for ciliochoroidal melanoma.

Methyl-CCNU, doxorubicin, and cis-diaminedichloroplatinum II in the management of recurrent and metastatic squamous carcinoma of the cervix

Twenty‐three patients with recurrent unresectable carcinoma of the cervix or distant metastasis at initial presentation were treated with methyl‐CCNU (175 mg/m2) and doxorubicin (45 mg/m2) on day 1 and Cis‐diaminedichloroplatinum II (90 mg/m2) on day 22 of a 42‐day treatment cycle. Twenty‐two patients had squamous carcinoma and 1 had adenosquamous carcinoma. There were two complete responses (CR), five partial responses (PR) (>50% tumor reduction, >3‐month duration), four patients with stable disease (<50% reduction, >3‐month duration), and 12 patients who had tumor progression. One CR has been maintained > 28 months, and the other >8 months. Total CR and PR was 7 of 23 (30.4%). Three responses occurred among 15 patients (20%) who had cancer primarily confined to the pelvis, while 4 of 8 patients (50%) with distant metastasis responded. During the initial 2 cycles of chemotherapy, 12 patients had myelosuppression, defined as a leukocytes <3000/mm3, granulocytes <1,000/mm3, or platelets <100,000/mm3. There were no treatment‐related deaths.

Association of chronic cystic mastopathy, xeromammographic patterns, and cancer.

The Breast Cancer Demonstration and Detection Project in Louisville (BCDDP‐L) screened 10,128 women for cancer. From this screening, another project evolved wherein those patients diagnosed as having chronic cystic mastopathy (CCM) were followed over a 10‐year period to evaluate any association between CCM and breast cancer. In all, 1396 breast biopsies were performed, with 165 cancers being diagnosed either on initial screening or during subsequent years. Three of these are excluded, since histopathologic slides could not be obtained for central review. Of this group, cancer was associated with CCM in 116 specimens and without CCM in 46 specimens. One subset of 355 patients with biopsy‐proven CCM but no breast cancer was followed for 6 to 12 years, for a total of 2443.5 woman‐years of observation. Within this subset, a total of only four cancers occurred (4 cancers/2443.5 woman‐years for 0.00164 cancers/woman‐years). This incidence is not significantly different from the expected value. However, an estimate is provided as to the power of the test that could be obtained from a larger sample size derived from other BCDDPs. This group of 355 women was sorted into subsets by establishing a matrix matching ten histopathologic subdivisions of CCM against six subdivisions of Wolfes xeromammographic (XM) patterns. The numbers of cancers in each cell of this matrix is reported. The results found no concentration of these four cancers in this matrix. Cancer 55:1372‐1375, 1985.

A stable linear algorithm for fitting the Lognormal model to survival data

The lognormal model can be fitted to survival data using a stable linear algorithm. When tested on 800 sets of mathematically generated data, this method proved more stable and efficient than the iterative method of maximum likelihood, which requires initial estimates of model parameters and failed to fit a substantial fraction of data sets. Though maximum likelihood yielded more consistent estimates of proportion cured, mean, and standard deviation of log(survival time), the linear normal algorithm may nevertheless prove useful for these purposes: (i) computing initial estimates of model parameters for the maximum likelihood method; (ii) fitting data sets that cannot be fit by this method; and (iii) deriving the lognormal model directly from cumulative mortality.

Validity of the clinical alert on breast cancer

In May 1988, the National Cancer Institute issued a clinical alert calling for the routine use of systemic adjuvant therapy for all node-negative breast cancers. Subsequent review of the data that the National Cancer Institute used as a basis for its endorsement revealed several limitations, including failure to consider cost-benefit ratios and failure to exclude late toxicities. The authors conclude that the issuance of the release was premature and that it does not attempt to balance the slight lengthening of disease-free survival against the overall population costs. It is suggested that physicians individually assess the potential merits of such a treatment regimen in each of their patients with node-negative breast cancer.

A method for assessing potential bias among cancer patients recorded as “dead of other causes”. Application to cases of intraocular melanoma

By applying the Cox regression model to 226 cases of intraocular melanoma, the authors detected a statistical association between tumor‐related measurements and 42 deaths that were recorded as due to causes other than melanoma within 10 years of follow‐up. This association may reflect a number of events, including errors in assignment of cause of death and confounding from any one of several sources.