A faster circular binary segmentation algorithm for the analysis of array CGH data.

MOTIVATION: Array CGH technologies enable the simultaneous measurement of DNA copy number for thousands of sites on a genome. We developed the circular binary segmentation (CBS) algorithm to divide the genome into regions of equal copy number. The algorithm tests for change-points using a maximal t-statistic with a permutation reference distribution to obtain the corresponding P-value. The number of computations required for the maximal test statistic is O(N2), where N is the number of markers. This makes the full permutation approach computationally prohibitive for the newer arrays that contain tens of thousands markers and highlights the need for a faster algorithm. RESULTS: We present a hybrid approach to obtain the P-value of the test statistic in linear time. We also introduce a rule for stopping early when there is strong evidence for the presence of a change. We show through simulations that the hybrid approach provides a substantial gain in speed with only a negligible loss in accuracy and that the stopping rule further increases speed. We also present the analyses of array CGH data from breast cancer cell lines to show the impact of the new approaches on the analysis of real data. AVAILABILITY: An R version of the CBS algorithm has been implemented in the "DNAcopy" package of the Bioconductor project. The proposed hybrid method for the P-value is available in version 1.2.1 or higher and the stopping rule for declaring a change early is available in version 1.5.1 or higher.

Properties of analysis methods that account for clustering in volume-outcome studies when the primary predictor is cluster size.

In recent years health services researchers have conducted 'volume-outcome' studies to evaluate whether providers (hospitals or surgeons) who treat many patients for a specialized condition have better outcomes than those that treat few patients. These studies and the inherent clustering of events by provider present an unusual statistical problem. The volume-outcome setting is unique in that 'volume' reflects both the primary factor under study and also the cluster size. Consequently, the assumptions inherent in the use of available methods that correct for clustering might be violated in this setting. To address this issue, we investigate via simulation the properties of three estimation procedures for the analysis of cluster correlated data, specifically in the context of volume-outcome studies. We examine and compare the validity and efficiency of widely-available statistical techniques that have been used in the context of volume-outcome studies: generalized estimating equations (GEE) using both the independence and exchangeable correlation structures; random effects models; and the weighted GEE approach proposed by Williamson et al. (Biometrics 2003; 59:36-42) to account for informative clustering. Using data generated either from an underlying true random effects model or a cluster correlated model we show that both the random effects and the GEE with an exchangeable correlation structure have generally good properties, with relatively low bias for estimating the volume parameter and its variance. By contrast, the cluster weighted GEE method is inefficient.

Outcome predictors for the increasing PSA state after definitive external-beam radiotherapy for prostate cancer.

PURPOSE: To identify predictors of distant metastases (DM) among patients who develop an isolated prostate-specific antigen (PSA) relapse after definitive external-beam radiotherapy for clinically localized prostate cancer. MATERIALS AND METHODS: A total of 1,650 patients with clinical stage T1 to T3 prostate cancer were treated with high-dose three-dimensional conformal radiotherapy. Of these, 381 patients subsequently developed three consecutive increasing PSA values and were characterized as having a biochemical relapse. The median follow-up time was 92 months from the completion of radiotherapy. RESULTS: The 5-year incidence of DM after an established PSA relapse was 29%. In a multivariate analysis, PSA doubling time (PSA-DT; P < .001), the clinical T stage (P < .001), and Gleason score (P = .007) were independent variables predicting for DM after established biochemical failure. The PSA-DT for favorable-, intermediate-, and unfavorable-risk patients who developed a biochemical failure was 20.0, 13.2, and 8.2 months, respectively (P < .001). The 3-year incidence of DM for patients with PSA-DT of 0 to 3, 3 to 6, 6 to 12, and more than 12 months was 49%, 41%, 20%, and 7%, respectively (P < .001). Patients with PSA-DT of 0 to 3 and 3 to 6 months demonstrated a 7.0 and 6.6 increased hazard of developing DM or death, respectively, compared with patients with a DT more than 12 months. CONCLUSION: In addition to clinical stage and Gleason score, PSA-DT was a powerful predictor of DM among patients who develop an isolated PSA relapse after external-beam radiotherapy for prostate cancer. Patients who develop biochemical relapse with PSA-DT < or = 6 months should be considered for systemic therapy or experimental protocols because of the high propensity for rapid DM development.

Circular binary segmentation for the analysis of array-based DNA copy number data.

DNA sequence copy number is the number of copies of DNA at a region of a genome. Cancer progression often involves alterations in DNA copy number. Newly developed microarray technologies enable simultaneous measurement of copy number at thousands of sites in a genome. We have developed a modification of binary segmentation, which we call circular binary segmentation, to translate noisy intensity measurements into regions of equal copy number. The method is evaluated by simulation and is demonstrated on cell line data with known copy number alterations and on a breast cancer cell line data set.

Two-stage designs for gene-disease association studies with sample size constraints.

Gene-disease association studies based on case-control designs may often be used to identify candidate polymorphisms (markers) conferring disease risk. If a large number of markers are studied, genotyping all markers on all samples is inefficient in resource utilization. Here, we propose an alternative two-stage method to identify disease-susceptibility markers. In the first stage all markers are evaluated on a fraction of the available subjects. The most promising markers are then evaluated on the remaining individuals in Stage 2. This approach can be cost effective since markers unlikely to be associated with the disease can be eliminated in the first stage. Using simulations we show that, when the markers are independent and when they are correlated, the two-stage approach provides a substantial reduction in the total number of marker evaluations for a minimal loss of power. The power of the two-stage approach is evaluated when a single marker is associated with the disease, and in the presence of multiple disease-susceptibility markers. As a general guideline, the simulations over a wide range of parametric configurations indicate that evaluating all the markers on 50% of the individuals in Stage 1 and evaluating the most promising 10% of the markers on the remaining individuals in Stage 2 provides near-optimal power while resulting in a 45% decrease in the total number of marker evaluations.

A method for evaluating the impact of individual haplotypes on disease incidence in molecular epidemiology studies.

Estimation of the association between haplotypes and disease from a case-control study is considered. Assuming a single "disease haplotype'' leads to the increased risk, attention focusses on the relative risks associated with a single copy, or two copies of the disease haplotype, relative to individuals with no copies. In this setting, case frequencies of the haplotype pairs are in Hardy-Weinberg Equilibrium (HWE) only if the combined influence of the two copies of the disease haplotype on risk is multiplicative. Thus, imputation cannot rely on the assumption of HWE for cases. A method is presented for obtaining estimates of the relative risks, making use of the EM algorithm and the assumption of HWE only for controls. The method accounts for the additional variation in the estimates due to the imputation of expected frequencies of haplotype pairs from ambiguous genotypes. A simulation study shows that the resulting confidence intervals have nominal coverage, and that the methods based on the assumption of HWE for both cases and controls can lead to bias.

High-dose intensity modulated radiation therapy for prostate cancer: early toxicity and biochemical outcome in 772 patients.

PURPOSE: To report the acute and late toxicity and preliminary biochemical outcomes in 772 patients with clinically localized prostate cancer treated with high-dose intensity-modulated radiotherapy (IMRT). METHODS AND MATERIALS: Between April 1996 and January 2001, 772 patients with clinically localized prostate cancer were treated with IMRT. Treatment was planned using an inverse-planning approach, and the desired beam intensity profiles were delivered by dynamic multileaf collimation. A total of 698 patients (90%) were treated to 81.0 Gy, and 74 patients (10%) were treated to 86.4 Gy. Acute and late toxicities were scored by the Radiation Therapy Oncology Group morbidity grading scales. PSA relapse was defined according to The American Society of Therapeutic Radiation Oncology Consensus Statement. The median follow-up time was 24 months (range: 6-60 months). RESULTS: Thirty-five patients (4.5%) developed acute Grade 2 rectal toxicity, and no patient experienced acute Grade 3 or higher rectal symptoms. Two hundred seventeen patients (28%) developed acute Grade 2 urinary symptoms, and one experienced urinary retention (Grade 3). Eleven patients (1.5%) developed late Grade 2 rectal bleeding. Four patients (0.1%) experienced Grade 3 rectal toxicity requiring either one or more transfusions or a laser cauterization procedure. No Grade 4 rectal complications have been observed. The 3-year actuarial likelihood of >/= late Grade 2 rectal toxicity was 4%. Seventy-two patients (9%) experienced late Grade 2 urinary toxicity, and five (0.5%) developed Grade 3 urinary toxicity (urethral stricture). The 3-year actuarial likelihood of >/= late Grade 2 urinary toxicity was 15%. The 3-year actuarial PSA relapse-free survival rates for favorable, intermediate, and unfavorable risk group patients were 92%, 86%, and 81%, respectively. CONCLUSIONS: These data demonstrate the feasibility of high-dose IMRT in a large number of patients. Acute and late rectal toxicities seem to be significantly reduced compared with what has been observed with conventional three-dimensional conformal radiotherapy techniques. Short-term PSA control rates seem to be at least comparable to those achieved with three-dimensional conformal radiotherapy at similar dose levels. Based on this favorable risk:benefit ratio, IMRT has become the standard mode of conformal treatment delivery for localized prostate cancer at our institution.

Two-stage designs for gene-disease association studies.

The goal of this article is to describe a two-stage design that maximizes the power to detect gene-disease associations when the principal design constraint is the total cost, represented by the total number of gene evaluations rather than the total number of individuals. In the first stage, all genes of interest are evaluated on a subset of individuals. The most promising genes are then evaluated on additional subjects in the second stage. This will eliminate wastage of resources on genes unlikely to be associated with disease based on the results of the first stage. We consider the case where the genes are correlated and the case where the genes are independent. Using simulation results, it is shown that, as a general guideline when the genes are independent or when the correlation is small, utilizing 75% of the resources in stage 1 to screen all the markers and evaluating the most promising 10% of the markers with the remaining resources provides near-optimal power for a broad range of parametric configurations. This translates to screening all the markers on approximately one quarter of the required sample size in stage 1.

Fitting tumor control probability models to biopsy outcome after three-dimensional conformal radiation therapy of prostate cancer: pitfalls in deducing radiobiologic parameters for tumors from clinical data.

PURPOSE: The goal of tumor control probability (TCP) models is to predict local control for inhomogeneous dose distributions. All existing fits of TCP models to clinical data have utilized summaries of dose distributions (e.g., prescription dose). Ideally, model fits should be based on dose distributions in the tumor, but usually only dose-volume histograms (DVH) of the planning target volume (PTV) are available. We fit TCP models to biopsy outcome after three-dimensional conformal radiation therapy of prostate cancer using either a dose distribution summary or the full DVH in the PTV. We discuss differences in the radiobiologic parameters and dose-response curves and demonstrate pitfalls in interpreting the results. METHODS AND MATERIAL: Two mechanistic TCP models were fit with a maximum likelihood technique to biopsy outcome from 103 prostate patients treated at Memorial Sloan-Kettering Cancer Center. Fits were performed separately for different patient subgroups defined by tumor-related prognostic factors. Fits were based both on full DVHs, denoted TCP(DVH(calc)), and, alternatively, assuming a homogeneous PTV dose given by the mean dose (Dmean) of each DVH, denoted TCP(Dmean(calc)). Dose distributions for these patients were very homogeneous with any cold spots located on the periphery of the PTV. These cold spots were uncorrelated with biopsy outcome, likely because the low-dose regions may not contain tumor cells. Therefore, fits of TCP models that are potentially sensitive to cold spots (e.g., TCP(DVH(calc))) likely give biologic parameters that diminish this sensitivity. In light of this, we examined differences in fitted clonogenic cell number, N(C), or density, rho(C), surviving fraction after 2 Gy, SF(2), or radiosensitivity, alpha, and their standard deviations in the population, sigma(SF(2)) and sigma(alpha), resulting from fits based on TCP(DVH(calc)) and TCP(Dmean(calc)). Dose-response curves for homogeneous irradiation (characterized by TCD(50), the dose for a TCP of 50%) and differences in TCP predictions calculated from the DVH using alternatively derived parameters were evaluated. RESULTS: Fits of TCP(Dmean(calc)) are better (i.e., have larger likelihood) than fits of TCP(DVH(calc)). For TCP(Dmean(calc)) fits, matching values of SF(2) and sigma(SF(2)) (or alpha and sigma(alpha)) exist for all N(C) (rho(C)) above a threshold that give fits of equal quality, with no maximum in likelihood. In contrast, TCP(DVH(calc)) fits have maximum likelihood for high SF(2) (low alpha) values that minimize effects of cold spots. Consequently, small N(C) (rho(C)) values are obtained to match the observed control rate. For example, for patients in low-, intermediate-, and high-risk groups, optimum values of SF(2) and N(C) are 0.771 and 3.3 x 10(3), 0.736 and 2.2 x 10(4), and 0.776 and 1.0 x 10(4), respectively. The TCD(50) of dose-response curves for intermediate-risk patients is 2.6 Gy lower using TCP(DVH(calc)) parameters (TCD(50) = 67.8 Gy) than for TCP(Dmean(calc)) parameters (TCD(50) = 70.4 Gy). TCP predictions calculated from the DVH using risk group-dependent TCP(Dmean(calc)) parameters are up to 53% lower than corresponding calculations with TCP(DVH(calc)) parameters. CONCLUSION: For our data, TCP parameters derived from DVHs likely do not reflect true radiobiologic parameters in the tumor, but are a consequence of the reduced importance of low-dose regions at the periphery of the PTV. Deriving radiobiologic parameters from TCP(Dmean(calc)) fits is not possible unless one parameter is already known. TCP predictions using TCP(DVH(calc)) and TCP(Dmean(calc)) parameters may differ substantially, requiring consistency in the derivation and application of model parameters. The proper derivation of radiobiologic parameters from clinical data requires both substantial dose inhomogeneities and understanding of how these coincide with tumor location.

Morbidity and mortality after neoadjuvant therapy for lung cancer: the risks of right pneumonectomy.

BACKGROUND: The risks of complications in patients undergoing thoracotomy after neoadjuvant therapy for nonsmall cell lung cancer remain controversial. We reviewed our experience to define it further. METHODS: All patients undergoing thoracotomy after induction chemotherapy from 1993 through 1999 were reviewed. Univariate and multivariate methods for logistic regression model were used to identify predictors of adverse events. RESULTS: Induction chemotherapy included mitomycin, vinblastine, and cisplatin (179 patients), carboplatin and paclitaxel (152 patients), and other combinations (139 patients). Eighty-five patients (18%) received preoperative radiation. Operations were pneumonectomy (97 patients), lobectomy (297 patients), lesser resection (18 patients), and exploration only (58 patients). Total mortality was 7 of 297 (2.4%) and 11 of 97 (11.3%) for all lobectomies and pneumonectomies, respectively, but mortality was 11 of 46 (23.9%) for right pneumonectomy. Complications developed in 179 patients (38%). By multiple regression analysis, right pneumonectomy (p = 0.02), blood loss (p = 0.01), and forced expiratory volume in one second (percent predicted) (p = 0.01) predicted complications. No factor emerged to explain this high right pneumonectomy mortality rate. CONCLUSIONS: Pulmonary resection after neoadjuvant therapy is associated with acceptable overall morbidity and mortality. However, right pneumonectomy is associated with a significantly increased risk and should be performed only in selected patients.

Identification of prognostic factors in advanced epithelial ovarian carcinoma.

OBJECTIVE: The Gynecologic Oncology Group (GOG) has demonstrated that age, tumor grade, and size and number of residual lesions after primary cytoreductive surgery are significant prognostic factors in advanced ovarian carcinoma. Recent studies have reported numerous other clinical features as having prognostic value. We sought to identify the independent prognostic factors for survival in a cohort of patients with advanced ovarian cancer. METHODS: We performed a retrospective chart review of all patients with stage III and IV ovarian carcinoma who received their primary treatment at our institution between 1987 and 1994. RESULTS: A total of 295 patients were identified, 282 of whom were evaluable. Of these 282 patients, 214 (76%) have died of disease or other causes. The median follow-up is 32 months (range: 1-139). Eighteen factors were evaluated for prognostic significance. Significant factors in univariate analysis included patient age, gravidity (0 vs > 0), parity (0 vs > 0), preoperative albumin level, preoperative total protein level, ascites (presence vs absence), disease stage (IIIA/IIIB vs IIIC vs IV), number of residual lesions (< or =20 vs >20), and diameter of largest residual tumor nodule (< or = 1 cm vs 1-2 cm vs > 2 cm). However, on multivariate analysis, only patient age (P < 0.001), ascites (P = 0.001), and size of residual disease (P = 0.005) retained prognostic significance. Substage of disease was of borderline significance (P = 0.086). CONCLUSION: Although numerous clinical variables have recently been reported to have prognostic value in advanced ovarian carcinoma, only patient age, presence or absence of ascites, and diameter of the largest residual tumor nodule proved to be of statistical significance in our analysis.

High dose radiation delivered by intensity modulated conformal radiotherapy improves the outcome of localized prostate cancer.

PURPOSE: We present the long-term outcome and tolerance of 3-dimensional (D) conformal and intensity modulated radiation therapy for localized prostate cancer. MATERIALS AND METHODS: Between October 1988 and December 1998, 1,100 patients with clinical stages T1c-T3 prostate cancer were treated with 3-D conformal or intensity modulated radiation therapy. Patients were categorized into prognostic risk groups based on pretreatment prostate specific antigen (PSA), Gleason score and clinical stage. Sextant biopsies were performed 2.5 years or greater after treatment to assess local control. PSA relapse was defined according to the consensus guidelines of the American Society for Therapeutic Radiation Oncology. Late toxicity was classified according to the Radiation Therapy Oncology Group morbidity grading scale. Median followup was 60 months. RESULTS: At 5 years the PSA relapse-free survival rate in patients at favorable, intermediate and unfavorable risk was 85% (95% confidence interval [CI] +/- 4), 58% (95% CI +/- 6) and 38% (95% CI +/- 6), respectively (p <0.001). Radiation dose was the most powerful variable impacting PSA relapse-free survival in each prognostic risk group. The 5-year actuarial PSA relapse-free survival rate for patients at favorable risk who received 64.8 to 70.2 Gy. was 77% (95% CI +/- 8) compared to 90% (95% CI +/- 8) for those treated with 75.6 to 86.4 Gy. (p = 0.04) [corrected]. The corresponding rates were 50% (95% CI +/- 8) versus 70% (95% CI +/- 6) in intermediate risk cases (p = 0.001), and 21% (95% CI +/- 8) versus 47% (95% CI +/- 6) in unfavorable risk cases (p = 0.008) [corrected]. Only 4 of 41 patients (10%) who received 81 Gy. had a positive biopsy 2.5 years or greater after treatment compared with 27 of 119 (23%) after 75.6, 23 of 68 (34%) after 70.2 and 13 of 24 (54%) after 64.8 Gy. The incidence of toxicity after 3-D conformal radiation therapy was dose dependent. The 5-year actuarial rate of grade 2 rectal toxicity in patients who received 75.6 Gy. or greater was 14% (95% CI +/- 2) compared with 5% (95% CI +/- 2) in those treated at lower dose levels (p <0.001). Treatment with intensity modulated radiation therapy significantly decreased the incidence of late grade 2 rectal toxicity since the 3-year actuarial incidence in 189 cases managed by 81 Gy. was 2% (95% CI +/- 2) compared with 14% (95% CI +/- 2) in 61 managed by the same dose of 3-D conformal radiation therapy (p = 0.005). The 5-year actuarial rate of grade 2 urinary toxicity in patients who received 75.6 Gy. or greater 3-D conformal radiation therapy was 13% compared with 4% in those treated up to lower doses (p <0.001). Intensity modulated radiation therapy did not affect the incidence of urinary toxicity. CONCLUSIONS: Sophisticated conformal radiotherapy techniques with high dose 3-D conformal and intensity modulated radiation therapy improve the biochemical outcome in patients with favorable, intermediate and unfavorable risk prostate cancer. Intensity modulated radiation therapy is associated with minimal rectal and bladder toxicity, and, hence, represents the treatment delivery approach with the most favorable risk-to-benefit ratio.

Late rectal bleeding after conformal radiotherapy of prostate cancer. II. Volume effects and dose-volume histograms.

PURPOSE AND OBJECTIVE: Late rectal bleeding is a potentially dose limiting complication of three-dimensional conformal radiotherapy (3D-CRT) for prostate cancer. The frequency of late rectal bleeding has been shown to increase as the prescription dose rises above 70 Gy. The purpose of this study is to identify features of the cumulative dose-volume histogram (DVH) for the rectal wall that correlate with late rectal bleeding after 3D-CRT for prostate cancer. METHODS AND MATERIALS: Follow-up information on rectal bleeding is available for 261 and 315 patients treated using 3D-CRT at Memorial Sloan-Kettering Cancer Center for Stage T1c-T3 prostate cancer with minimum target doses of 70.2 and 75.6 Gy, respectively. All patients in this study were treated with a coplanar 6-field technique (2 lateral and 4 oblique fields). Patients were classified as having rectal bleeding if they bled (> or = Grade 2) before 30 months, and nonbleeding (< or = Grade 1) if they were without bleeding at 30 months, using the RTOG morbidity scale. Rectal bleeding was observed in 13 and 38 of the patients treated at 70.2 and 75.6 Gy, respectively. Treatment plans were analyzed for 39 nonbleeding and 13 bleeding patients receiving 70.2 Gy, and 83 nonbleeding and 36 bleeding patients receiving 75.6 Gy. Dose-volume histograms (DVHs) for the anatomic rectal wall were calculated. Average DVHs of the bleeding and nonbleeding patients were generated, and a permutation test was used to assess the significance of differences between them, for each dose group. The confounding effect of total rectal wall volume (V(RW)) was removed by calculating the average differences in DVHs between all combinations of bleeding and nonbleeding patients with similar V(RW)s. Finally, multivariate analysis using logistic regression was performed to test the significance of the DVH variables in the presence of anatomic, geometric, and medical variables previously found to correlate with rectal bleeding in a companion analysis of the same patients. RESULTS: The area under the average percent volume DVH for the rectal wall of patients with bleeding was significantly higher than those of patients without bleeding in both dose groups (p = 0.02, 70.2 Gy; p < 0.0001, 75.6 Gy). However, small V(RW)s were associated with rectal bleeding (p = 0.06, 70.2 Gy; p < 0.01, 75.6 Gy), resulting in an increase in average percent volumes exposed to all doses for patients with rectal bleeding. For patients with similar V(RW)s, rectal bleeding was significantly correlated with the volumes exposed to 46 Gy in both dose groups (p = 0.02, 70.2 Gy; p = 0.005, 75.6 Gy, tolerance in V(RW): 5 ccs). For the 75.6 Gy dose group, the percent volume receiving 77 Gy was significantly correlated with rectal bleeding (p < 0.005). Bivariate analysis using logistic regression, including V(RW) together with a single DVH variable, showed good agreement with the above analysis. Multivariate analysis revealed a borderline significant correlation of the percent volume receiving 71 Gy in the 70.2 Gy dose group. It also showed that the DVH variables were highly correlated with geometric and dosimetric variables previously found to correlate with rectal bleeding in multivariate analysis. CONCLUSION: Significant volume effects were found in the probability of late rectal bleeding for patients undergoing 3D-CRT for prostate cancer with prescription doses of 70.2 and 75.6 Gy. The percent volumes exposed to 71 and 77 Gy in the 70.2 and 75.6 Gy dose groups respectively were significantly correlated with rectal bleeding. The independent correlation of small V(RW) with rectal bleeding may indicate the existence of a functional reserve for the rectum. The independent association with larger percent volumes exposed to intermediate doses ( approximately 46 Gy) seen in both dose groups may indicate that a large surrounding region of intermediate dose may interfere with the ability to repair the effects of a central high dose region.

A permutation test to compare receiver operating characteristic curves.

We developed a permutation test in our earlier paper (Venkatraman and Begg, 1996, Biometrika 83, 835-848) to test the equality of receiver operating characteristic curves based on continuous paired data. Here we extend the underlying concepts to develop a permutation test for continuous unpaired data, and we study its properties through simulations.

Postimplantation dosimetric analysis of permanent transperineal prostate implantation: improved dose distributions with an intraoperative computer-optimized conformal planning technique.

PURPOSE: To compare the target coverage and dose to normal tissues after I-125 transperineal permanent implantation (TPI) of the prostate in 90 patients treated with one of three different transperineal techniques. METHODS AND MATERIALS: Detailed postimplant dosimetric evaluations of permanent I-125 implantation procedures were performed on 30 consecutive patients treated between 1995-1996 who underwent TPI using a preplanning CT-based technique, on 30 consecutive patients treated in 1997-1998 who underwent an ultrasound-guided approach with intraoperative determination of seed distribution based on an I-125 nomogram, and on 30 consecutive patients in 1998-1999 who underwent TPI with intraoperative computer-based 3-dimensional conformal optimization. For all three techniques, postimplant CT scans were obtained 4-6 hours after TPI. Dosimetric parameters included V(100), V(90), V(150), D(100), D(90), D(80), as well as maximal and average doses to the urethra and rectal wall. These parameter outcomes are reported as a percentage of the prescription dose. RESULTS: The intraoperative 3D-optimized technique (I-3D) provided superior target coverage with the prescription dose for all dosimetric variables evaluated compared to the other treatment techniques. The median V(100), V(90), and D(90) values for the I-3D technique were 96%, 98%, and 116%, respectively. In contrast, the V(100), V(90), and D(90) values for the CT preplan and ultrasound manual optimization approaches were 86%, 89%, and 88%, respectively and 88%, 92%, and 94%, respectively (I-3D versus other techniques: p < 0.001). The superior target coverage with the I-3D technique was also associated with a higher cumulative implant activity required by the optimization program. A multivariate analysis determined that the treatment technique (I-3D versus other approaches) was an independent predictor of improved target coverage for each parameter analyzed (p < 0.001). In addition, higher cumulative implant activities and smaller prostate target volumes were independent predictors of improved target coverage. The maximum and average urethral doses were significantly lower with the I-3D technique compared to the other techniques; a modest increase in the average rectal dose was also observed with this approach. CONCLUSION: Three-dimensional intraoperative computer optimized TPI consistently provided superior target coverage with the prescription dose and significantly lower urethral doses compared to two other techniques used. These data provide proof-of-principle that improved therapeutic ratios can be achieved with the integration of more sophisticated intraoperative planning for TPI and may potentially have a profound impact on the outcome of patients treated with this modality.

Comparing tumour staging and grading systems: a case study and a review of the issues, using thymoma as a model.

We consider the problem of comparing alternative cancer staging and grading systems. Statistical comparisons are on the basis of the ability to predict survival, but more qualitative criteria, such as parsimony, and distinctive prognostic separability of the categories are relevant also. Furthermore, some staging systems are clearly ordinal, while others are not. Three candidate statistical measures are studied and compared: explained variation; area under the ROC curve; and the probability of concordance of stage and survival. Each of these has individual strengths and weaknesses. A data set involving the staging of thymoma is analysed in detail to motivate the problem and illustrate the results.

Analysis of biopsy outcome after three-dimensional conformal radiation therapy of prostate cancer using dose-distribution variables and tumor control probability models.

PURPOSE: To investigate tumor control following three-dimensional conformal radiation therapy (3D-CRT) of prostate cancer and to identify dose-distribution variables that correlate with local control assessed through posttreatment prostate biopsies. METHODS AND MATERIAL: Data from 132 patients, treated at Memorial Sloan-Kettering Cancer Center (MSKCC), who had a prostate biopsy 2.5 years or more after 3D-CRT for T1c-T3 prostate cancer with prescription doses of 64.8-81 Gy were analyzed. Variables derived from the dose distribution in the PTV included: minimum dose (Dmin), maximum dose (Dmax), mean dose (Dmean), dose to n% of the PTV (Dn), where n = 1%,...,99%. The concept of the equivalent uniform dose (EUD) was evaluated for different values of the surviving fraction at 2 Gy (SF(2)). Four tumor control probability (TCP) models (one phenomenologic model using a logistic function and three Poisson cell kill models) were investigated using two sets of input parameters, one for low and one for high T-stage tumors. Application of both sets to all patients was also investigated. In addition, several tumor-related prognostic variables were examined (including T-stage, Gleason score). Univariate and multivariate logistic regression analyses were performed. The ability of the logistic regression models (univariate and multivariate) to predict the biopsy result correctly was tested by performing cross-validation analyses and evaluating the results in terms of receiver operating characteristic (ROC) curves. RESULTS: In univariate analysis, prescription dose (Dprescr), Dmax, Dmean, dose to n% of the PTV with n of 70% or less correlate with outcome (p < 0.01). The area under the ROC curve for Dmean is 0.64. In contrast, Dmin (p = 0.6), D98 (p = 0.2) or D95 (p = 0.1) are not significantly correlated with outcome. The results for EUD depend on the input parameter SF(2): EUD correlates significantly with outcome for SF(2) of 0.4 or more, but not for lower SF(2) values. Using either of the two input parameters sets, all TCP models correlate with outcome (p < 0.05; ROC areas 0.60-0.62). Using T-stage dependent input parameters, the correlation is improved (logistic function: p < 0.01, ROC area 0.67, Poisson models: p < 0.01, ROC areas 0.64-0.66). In comparison, the ROC area is 0.68 for the combination of Dmean and T-stage. After multivariate analysis, a model based on TCP, D20 and Gleason score is the best overall model (ROC area 0.73). However, an alternative model based on Dmean, Gleason score, and T-stage is competitive (ROC area 0.70). CONCLUSION: Biopsy outcome after 3D-CRT of prostate cancer at MSKCC is not correlated with Dmin in the PTV and appears to be insensitive to cold spots in the dose distribution. This observation likely reflects the fact that much of the PTV, especially at the periphery, may not contain viable tumor cells and that the treatment margins were sufficiently large. Therefore, the predictive power of all variables which are sensitive to cold spots, like TCPs with Poisson models and EUD for low SF(2), is limited because the low dose region may not coincide with the tumor location. Instead, for MSKCC prostate cancer patients with their standardized CTV definition, substantial target motion and small dose inhomogeneities, Dmean (or any variable that downplays the effect of cold spots) is a very good predictor of biopsy outcome. While our findings may indicate a general problem in the application of current TCP models to clinical data, these conclusions should not be extrapolated to other disease sites without careful analysis.

Clinical experience with intensity modulated radiation therapy (IMRT) in prostate cancer.

PURPOSE: To compare acute and late toxicities of high-dose radiation for prostate cancer delivered by either conventional three-dimensional conformal radiation therapy (3D-CRT) or intensity modulated radiation therapy (IMRT). MATERIALS AND METHODS: Between September 1992 and February 1998, 61 patients with clinical stage T1c- T3 prostate cancer were treated with 3D-CRT and 171 with IMRT to a prescribed dose of 81 Gy. To quantitatively evaluate the differences between conventional 3D-CRT and IMRT, 20 randomly selected patients were planned concomitantly by both techniques and the resulting treatment plans were compared. Acute and late radiation-induced morbidity was evaluated in all patients and graded according to the Radiation Therapy Oncology Group toxicity scale. RESULTS: Compared with conventional 3D-CRT, IMRT improved the coverage of the clinical target volume (CTV) by the prescription dose and reduced the volumes of the rectal and bladder walls carried to high dose levels (P<0.01), indicating improved conformality with IMRT. Acute and late urinary toxicities were not significantly different for the two methods. However, the combined rates of acute grade 1 and 2 rectal toxicities and the risk of late grade 2 rectal bleeding were significantly lower in the IMRT patients. The 2-year actuarial risk of grade 2 bleeding was 2% for IMRT and 10% for conventional 3D-CRT (P<0.001). CONCLUSIONS: The data demonstrate the feasibility and safety of high-dose IMRT for patients with localized prostate cancer and provide a proof-of-principle that this method improves dose conformality relative to tumor coverage and exposure to normal tissues.

Late rectal toxicity after conformal radiotherapy of prostate cancer (I): multivariate analysis and dose-response.

PURPOSE: The purpose of this paper is to use the outcome of a dose escalation protocol for three-dimensional conformal radiation therapy (3D-CRT) of prostate cancer to study the dose-response for late rectal toxicity and to identify anatomic, dosimetric, and clinical factors that correlate with late rectal bleeding in multivariate analysis. METHODS AND MATERIALS: Seven hundred forty-three patients with T1c-T3 prostate cancer were treated with 3D-CRT with prescribed doses of 64.8 to 81.0 Gy. The 5-year actuarial rate of late rectal toxicity was assessed using Kaplan-Meier statistics. A retrospective dosimetric analysis was performed for patients treated to 70.2 Gy (52 patients) or 75.6 Gy (119 patients) who either exhibited late rectal bleeding (RTOG Grade 2/3) within 30 months after treatment (i.e., 70.2 Gy-13 patients, 75. 6 Gy-36 patients) or were nonbleeding for at least 30 months (i.e., 70.2 Gy-39 patients, 75.6 Gy-83 patients). Univariate and multivariate logistic regression was performed to correlate late rectal bleeding with several anatomic, dosimetric, and clinical variables. RESULTS: A dose response for >/= Grade 2 late rectal toxicity was observed. By multivariate analysis, the following factors were significantly correlated with >/= Grade 2 late rectal bleeding for patients prescribed 70.2 Gy: 1) enclosure of the outer rectal contour by the 50% isodose on the isocenter slice (i.e., Iso50) (p < 0.02), and 2) smaller anatomically defined rectal wall volume (p < 0.05). After 75.6 Gy, the following factors were significant: 1) smaller anatomically defined rectal wall volume (p < 0.01), 2) higher rectal D(max) (p < 0.01), 3) enclosure of rectal contour by Iso50 (p < 0.01), 4) patient age (p = 0.02), and 5) history of diabetes mellitus (p = 0.04). In addition to these five factors, acute rectal toxicity was also significantly correlated (p = 0.05) with late rectal bleeding when patients from both dose groups were combined in multivariate analysis. CONCLUSION: A multivariate logistic regression model is presented which describes the probability of developing late rectal bleeding after conformal irradiation of prostate cancer. Late rectal bleeding correlated with factors which may indicate that a greater fractional volume of rectal wall was exposed to high dose, such as smaller rectal wall volume, inclusion of the rectum within the 50% isodose on the isocenter slice, and higher rectal D(max).

The ability of preoperative serum CA-125 to predict optimal primary tumor cytoreduction in stage III epithelial ovarian carcinoma.

PURPOSE: The aim of this study was to determine the ability of preoperative serum CA-125 to predict optimal primary tumor cytoreduction in patients with Stage III epithelial ovarian carcinoma. METHODS: We performed a retrospective chart review of 100 consecutive patients with Stage III ovarian carcinoma who had a serum CA-125 drawn prior to primary cytoreductive surgery. We used a receiver operating characteristic curve to determine the CA-125 level with the maximal prognostic power in predicting optimal versus suboptimal cytoreduction. RESULTS: The median CA-125 level for the 100 patients was 819 U/ml (range 5.6-26,200 U/ml). Optimal cytoreduction (diameter of largest residual tumor nodule < or =1 cm) was obtained in 45 cases (45%). The probability of performing optimal cytoreduction decreased with increasing CA-125 levels. A preoperative CA-125 level of 500 U/ml was identified as the value with the most predictive power. Optimal cytoreduction was achieved in 33 of the 45 cases (73%) with a CA-125 less than 500 U/ml compared to only 12 of the 55 cases (22%) with a CA-125 greater than 500 U/ml. Using a threshold level of 500 U/ml, the preoperative serum CA-125 level was able to predict optimal versus suboptimal cytoreduction with a sensitivity of 78%, specificity of 73%, positive predictive value of 78%, and negative predictive value of 73%. CONCLUSION: The probability of performing optimal cytoreduction in patients with Stage III ovarian carcinoma and a preoperative CA-125 greater than 500 U/ml was approximately one in five. These patients may be candidates for initial laparoscopic evaluation to obtain a confirmatory tissue diagnosis and to determine resectability.