Categorizing a prognostic variable: review of methods, code for easy implementation and applications to decision-making about cancer treatments.

Categorizing prognostic variables is essential for their use in clinical decision-making. Often a single cutpoint that stratifies patients into high-risk and low-risk categories is sought. These categories may be used for making treatment recommendations, determining study eligibility, or to control for varying patient prognoses in the design of a clinical trial. Methods used to categorize variables include: biological determination (most desirable but often unavailable); arbitrary selection of a cutpoint at the median value; graphical examination of the data for a threshold effect; and exploration of all observed values for the one which best separates the risk groups according to a chi-squared test. The last method, called the minimum p-value approach, involves multiple testing which inflates the type I error rates. Several methods for adjusting the inflated p-values have been proposed but remain infrequently used. Exploratory methods for categorization and the minimum p-value approach with its various p-value corrections are reviewed, and code for their easy implementation is provided. The combined use of these methods is recommended, and demonstrated in the context of two cancer-related examples which highlight a variety of the issues involved in the categorization of prognostic variables.

Complete percutaneous excision of infiltrating carcinoma at stereotactic breast biopsy: how can tumor size be assessed?

OBJECTIVE: The purpose of this study was to determine the frequency of complete excision of infiltrating carcinoma at stereotactic 11-gauge directional vacuum-assisted breast biopsy and to evaluate the feasibility of measuring tumor size in stereotactic biopsy specimens in infiltrating carcinomas that were percutaneously excised. MATERIALS AND METHODS: We performed retrospective review of 51 infiltrating carcinomas diagnosed using stereotactic 11-gauge directional vacuum-assisted biopsy that underwent subsequent surgery. For lesions yielding no residual infiltrating carcinoma at surgery, the maximal dimension of the tumor was measured in stereotactic biopsy specimens using ocular micrometry. RESULTS: In 10 (20%) (95% confidence intervals, 9.8-33.1%) of 51 infiltrating carcinomas diagnosed at stereotactic biopsy, surgery revealed no residual infiltrating carcinoma. Complete excision of infiltrating carcinoma was more frequent if 14 or more specimens were obtained (32% versus 0%, p < .004), if the mammographic lesion was removed (35% versus 7%, p < .03), and if the mammographic lesion size measured 0.7 cm or less (50% versus 16%, p = .08). Tumor size in stereotactic biopsy specimens was within 3 mm of mammographic lesion size in six (60%) of 10 lesions, including five (71%) of seven masses and one (33%) of three calcification lesions, but was smaller than the mammographic lesion size in eight (80%) of 10 lesions. CONCLUSION: Surgery revealed no residual infiltrating carcinoma in 10 (20%) of 51 infiltrating carcinomas diagnosed at stereotactic 11-gauge biopsy. Although tumor size can be assessed in stereotactic biopsy specimens in these lesions, such measurements may underestimate the maximal dimension of the tumor. Further study is needed to evaluate the usefulness of these measurements in guiding treatment decisions.

Ifosfamide, carboplatin, and etoposide: a highly effective cytoreduction and peripheral-blood progenitor-cell mobilization regimen for transplant-eligible patients with non-Hodgkin's lymphoma.

PURPOSE: To evaluate a chemotherapy regimen that consisted of ifosfamide administered as an infusion with bolus carboplatin, and etoposide (ICE) supported by granulocyte colony-stimulating factor (G-CSF) for cytoreduction and stem-cell mobilization in transplant-eligible patients with primary refractory or relapsed non-Hodgkin's lymphoma (NHL). PATIENTS AND METHODS: One hundred sixty-three transplant-eligible patients with relapsed or primary refractory NHL were treated from October 1993 to December 1997 with ICE chemotherapy at Memorial Sloan-Kettering Cancer Center. Administration of three cycles of ICE chemotherapy was planned at 2-week intervals. Peripheral-blood progenitor cells were collected after cycle 3, and all patients who achieved a partial response (PR) or complete response (CR) to ICE chemotherapy were eligible to proceed to transplantation. Event-free and overall survival, ICE-related toxicity, and the number of CD34(+) cells collected after treatment with ICE and G-CSF were evaluated. RESULTS: All 163 patients were assessable for response, and there was no treatment-related mortality. A major response (CR/PR) was evident in 108 patients (66.3%); 89% of the responding patients underwent successful transplantation. Patient who underwent transplantation and achieved a CR to ICE had a superior overall survival to that of patients who achieved a PR (65% v 30%; P =.003). The median number of CD34(+) cells/kg collected was 8.4 x 10(6). The dose-limiting toxicity of ICE was hematologic, with 29.4% of patients developing grade 3/4 thrombocytopenia. There were minimal nonhematologic side effects. CONCLUSION: ICE chemotherapy, with ifosfamide administered as a 24-hour infusion to decrease CNS side effects, and the substitution of carboplatin for cisplatin to minimize nephrotoxicity, is a very effective cytoreduction and mobilization regimen in patients with NHL. Furthermore, the quality of the clinical response to ICE predicts for posttransplant outcome.

The International Prognostic Index predicts for outcome following autologous stem cell transplantation in patients with relapsed and primary refractory intermediate-grade lymphoma.

We analyzed a group of 51 patients with primary refractory and relapsed intermediate-grade lymphoma (IGL) from the time of initiation of three cycles of second-line chemotherapy, ifosfamide, carboplatin and etoposide (ICE), in whom the intent was to administer curative high-dose chemoradiotherapy and autologous stem cell transplantation (ASCT). We sought to determine if the International Prognostic Index (IPI) assessed immediately prior to ICE, second-line IPI (sIPI), was predictive of outcome. The response rate to ICE-based chemotherapy was 69%, and 47% of the transplanted patients remain failure-free at 2.5 years. Stratification of patients based upon the sIPI demonstrated a superior 2.5 year failure-free survival (FFS) curve for patients with low (I) or low-intermediate (II) risk disease vs. those with high-intermediate (III) and high (IV) risk disease (45% vs. 9%, P<0.001). When the analysis was restricted to those patients with chemosensitive disease, the sIPI (I/II vs. III/IV) also separated patients into two distinct prognostic groups (59% vs. 20%, P = 0.04). Patients with sIPI I and II disease have a favorable outcome with ICE chemotherapy and ASCT. However, patients with sIPI III and IV disease derive limited benefit from this treatment strategy, and new approaches are needed in this patient group.

The breast imaging reporting and data system: positive predictive value of mammographic features and final assessment categories.

OBJECTIVE: The purpose of the study was to assess the positive predictive value of mammographic features and final assessment categories described in the Breast Imaging Reporting and Data System (BI-RADS) for lesions on which biopsies have been performed. SUBJECTS AND METHODS: We prospectively evaluated 492 impalpable mammographically detected lesions on which surgical biopsy (as opposed to percutaneous biopsy) was performed. Each lesion was classified according to BI-RADS descriptors for masses (margins and shape) and calcifications (morphology and distribution) and was categorized by the BI-RADS final assessment categories as category 3 ("probably benign"), category 4 ("suspicious abnormality"), or category 5 ("highly suggestive of malignancy"). Mammographic and pathologic findings were reviewed. RESULTS: Carcinoma was present in 225 (46%) of 492 lesions. For the 492 lesions subject to biopsy, BI-RADS final assessment categories were category 3 in eight lesions (2%), category 4 in 355 (72%), and category 5 in 129 (26%). The features with highest positive predictive value for carcinoma were spiculated margins (81%), irregular shape (73%), linear calcification morphology (81%), and segmental or linear calcification distribution (74% and 68%, respectively). Carcinoma was present in 105 (81%) of 129 category 5 lesions compared with 120 (34%) of 355 category 4 lesions (p < .001). The frequency of carcinoma was higher in category 5 than in category 4 lesions for all mammographic lesion types and all interpreting radiologists. CONCLUSION: The standardized terminology of the BI-RADS lexicon allows quantification of the likelihood of carcinoma in an impalpable breast lesion. The features with highest positive predictive value--spiculated margins, irregular shape, linear morphology, and segmental or linear distribution--warrant designation of a lesion as category 5.

Factors affecting mobilization of peripheral blood progenitor cells in patients with lymphoma.

The objective of this study was to identify factors associated with poor mobilization of peripheral blood progenitor cells (PBPCs) or delayed platelet engraftment after high-dose therapy and autologous stem cell transplantation in patients with lymphoma. Fifty-eight patients with Hodgkin's disease or non-Hodgkin's lymphoma underwent PBPC transplantation as the "best available therapy" at Memorial Sloan-Kettering Cancer Center (New York, NY) between 1993 and 1995. PBPCs were mobilized with either granulocyte colony-stimulating factor (G-CSF) alone (n = 19) or G-CSF following combination chemotherapy (n = 39). Forty-eight of these patients underwent a PBPC transplant, receiving a conditioning regimen containing cyclophosphamide, etoposide, and either total body irradiation, total lymphoid irradiation, or carmustine. A median number of 4.6 x 10(6) CD34+ cells/kg were obtained with a median of three leukapheresis procedures. Mobilization of PBPCs using chemotherapy plus G-CSF was superior to G-CSF alone (6.7 x 10(6) versus 1.5 x 10(6) CD34+ cells/kg; P = 0.0002). Poorer mobilization of progenitor cells was observed in patients who had previously received stem cell-toxic chemotherapy, including (a) nitrogen mustard, procarbazine, melphalan, carmustine or > 7.5 g of cytarabine chemotherapy premobilization (2.0 x 10(6) versus 6.0 x 10(6) CD34+ cells/kg; P = 0.005), or (b) > or = 11 cycles of any previous chemotherapy (2.6 x 10(6) versus 6.7 x 10(6) CD34+ cells/kg; P = 0.02). Platelet recovery to > 20,000/microliter was delayed in patients who received < 2.0 x 10(6) CD34+ cells (median, 13 versus 22 days; P = 0.06). Patients who received > or = 11 cycles of chemotherapy prior to PBPC mobilization tended to have delayed platelet recovery to > 20,000/microliter and to require more platelet transfusions than less extensively pretreated patients (median, 13.5 versus 23.5 days; P = 0.15; median number of platelet transfusion episodes, 13 versus 9; P = 0.17). These data suggest that current strategies to mobilize PBPCs may be suboptimal in patients who have received either stem cell-toxic chemotherapy or > or = 11 cycles of chemotherapy prior to PBPC mobilization. Alternative approaches, such as ex vivo expansion or the use of other growth factors in addition to G-CSF, may improve mobilization of progenitor cells for PBPC transplantation.

Analysis of cancers not diagnosed at stereotactic core breast biopsy.

PURPOSE: To evaluate the mammographic and histopathologic features of carcinomas not diagnosed at stereotactic core biopsy. MATERIALS AND METHODS: A retrospective review revealed 144 surgically confirmed carcinomas preoperatively sampled with stereotactic core biopsy. Diagnosis at stereotactic core biopsy was carcinoma in 116 (81%) lesions, atypical hyperplasia in 21 (15%), and benign findings discordant with those from mammography in seven (5%). Mammographic and histopathologic findings in the latter 28 cases were reviewed. RESULTS: Prompt repeat biopsy was recommended in all 28 cases. The frequency with which a cancer yielded atypical hyperplasia at stereotactic core biopsy was higher for calcifications than masses (30% vs 5%, P < .0001), ductal carcinoma in situ (DCIS) than infiltrating carcinoma (33% vs 7%, P = .0002), and noncomedo than comedo DCIS (60% vs 9%, P = .0008). No significant difference was observed in the likelihood of benign core biopsy findings without atypia in malignant calcifications versus masses (7% vs 3%, P = .43), DCIS versus infiltrating carcinoma (7% vs 4%, P = .43), or noncomedo versus comedo DCIS (0% vs 9%, P = .49). CONCLUSION: The likelihood of not diagnosing carcinoma was highest for calcifications and for noncomedo DCIS. Discordance in mammographic and histopathologic findings or the presence of atypical hyperplasia may enable the radiologist to identify missed or underestimated carcinomas prospectively and avoid a delay in diagnosis.