Portal de Periódicos da CAPES

Image analysis combined with visual cytology in the early detection of recurrent bladder carcinoma

1998; Wiley; Volume: 82; Issue: 9 Linguagem: Inglês

10.1002/(sici)1097-0142(19980501)82

1097-0142

Allison M. Richman, Susan T. Mayne, James F. Jekel, Peter C. Albertsen,

Esophageal Cancer Research and Treatment

CancerVolume 82, Issue 9 p. 1738-1748 Original ArticleFree Access Image analysis combined with visual cytology in the early detection of recurrent bladder carcinoma Allison M. Richman M.P.H., Corresponding Author Allison M. Richman M.P.H. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, Connecticut1600 N. Rhodes St., Arlington, VA 22209===Search for more papers by this authorSusan T. Mayne Ph.D., Susan T. Mayne Ph.D. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, ConnecticutSearch for more papers by this authorJames F. Jekel M.D., James F. Jekel M.D. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, ConnecticutSearch for more papers by this authorPeter Albertsen M.D., M.S., Peter Albertsen M.D., M.S. Division of Urology, Department of Surgery, University of Connecticut Health Center, Farmington, ConnecticutSearch for more papers by this author Allison M. Richman M.P.H., Corresponding Author Allison M. Richman M.P.H. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, Connecticut1600 N. Rhodes St., Arlington, VA 22209===Search for more papers by this authorSusan T. Mayne Ph.D., Susan T. Mayne Ph.D. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, ConnecticutSearch for more papers by this authorJames F. Jekel M.D., James F. Jekel M.D. Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, ConnecticutSearch for more papers by this authorPeter Albertsen M.D., M.S., Peter Albertsen M.D., M.S. Division of Urology, Department of Surgery, University of Connecticut Health Center, Farmington, ConnecticutSearch for more papers by this author First published: 31 October 2000 https://doi.org/10.1002/(SICI)1097-0142(19980501)82:9 3.0.CO;2-3Citations: 12AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract BACKGROUND Early detection of recurrent transitional cell carcinoma of the bladder (TCC) is important to permit early treatment, which produces maximal preservation of the bladder and maximum survival. METHODS This retrospective cohort study attempted to determine the period of time over which urinary DNA image analysis combined with visual cytology is useful in the early detection of recurrent TCC of the bladder. The authors believe this study is unique in that it measured the effectiveness of this test (image analysis plus visual cytology combined) at varying times before clinical diagnosis of recurrence was made. The cohort was comprised of 175 urologic patients from urologic practices across the U.S. Data, collected between January 1991 and February 1994, included cystoscopy, biopsy, DNA image analysis, and visual cytologic reports. RESULTS Sixty patients in the cohort were found to have active TCC whereas 115 patients had a history of, but no active, disease during the follow-up period. As expected, the sensitivity and specificity of DNA image analysis in combination with visual cytology, and DNA image analysis alone, were greatest when urinary samples were obtained close to the time of diagnosis. In general, the longer the interval from the combined tests to the time of diagnosis, the lower the sensitivity. The combined tests had predictive value up to 3 months prior to clinical diagnosis when any detectable cytologic abnormality was considered positive. At the optimal cutoff points as determined from receiver operating characteristic curves, sensitivity increased when DNA image analysis was supplemented with visual cytology. CONCLUSIONS The combination of DNA image analysis and visual cytology provides a better method for the early detection of recurrent TCC than DNA image analysis alone. This test potentially may be useful in providing information regarding bladder tumor recurrence up to 3 months prior to clinical evidence of disease. Cancer 1998;82:1738-48. © 1998 American Cancer Society. Superficial tumors of the bladder recur locally in approximately 50% of transitional cell carcinoma (TCC) patients within 6-12 months of their removal.1 Therefore, patients with TCC are at high risk for repeated tumor episodes. If left undetected and untreated, such tumors may progress and require more radical and debilitating procedures than if detected and treated early. The availability of an effective and noninvasive method for monitoring patients with TCC of the bladder would increase the lead time for treatment and improve outcomes. In addition to predicting disease recurrence, biomarkers also may be used as surrogate intermediate endpoints for monitoring chemopreventive therapy. DNA image analysis has received attention as a new, highly technical method for detecting and providing prognostic information regarding TCC of the bladder. 2-18 A more traditional way of measuring DNA content of bladder cells is flow cytometry.19-26 Studies comparing the value of image analysis with flow cytometry have shown varying results.27, 28 Several recent studies on image analysis have used bladder tissue specimens.2-12 Fewer studies have attempted to evaluate the use of image analysis for the detection and prognosis of bladder carcinoma using urine specimens.13-16 Cross-sectional studies have suggested that image analysis is more effective in identifying tumors than is visual cytology of urine specimens.13, 14 In particular, sensitivity is increased. Follow-up studies demonstrated through regression analysis that DNA image analysis can predict TCC progression, recurrence free survival, and overall survival.2-4 Gerber et al. found that DNA image analysis produced an improved sensitivity (93% vs. 36%) but decreased specificity (41% vs. 95%) compared with visual cytology.15 However, when both procedures were used together, both measures attained the highest level. Amberson and Laino reported improved sensitivity for DNA image analysis combined with visual cytology compared with visual cytology alone (84% vs. 72%).16 Based on reports that sensitivity is improved when DNA image analysis is combined with visual cytology of urine specimens,15, 16 the objective of the current study was to attempt to replicate those findings within a retrospective cohort design, to evaluate changes in sensitivity and specificity over time using mostly noninvasively obtained urine specimens, and to determine the length of time for which the combined tests have clinical utility. METHODS Subjects were selected from the data base of urologic patients throughout the U.S. who had urine samples analyzed by one diagnostic laboratory. The laboratory data base was used to identify all patients for whom DNA image analysis in combination with visual cytology had been performed. Two mailings were done to identify patients eligible for inclusion in the study and to collect relevant clinical data. The first mailing was sent to all urologists in the laboratory's data base meeting the following criteria: 1) the urologist began using the laboratory test (image analysis plus visual cytology) for urine specimens prior to December 31, 1990, and 2) the urologist had at least one patient who had had at least three independent urine samples tested. The mailing inquired as to whether the patient had a history of TCC and, if so, whether the TCC had recurred (i.e., showed a positive biopsy or positive clinical diagnosis) during the study period. The responses were used to exclude patients for whom the laboratory test was being used for reasons other than monitoring TCC and to obtain preliminary estimates on the rate of recurrence in this population. The second mailing included the data collection form used as a survey instrument for the physician. It was sent to the 113 urologists who reported having eligible patients in the first mailing. All eligible patients were potential study subjects. A total of 175 urologic patients, under the care of 50 private urologists, were studied. Recurrent patients had at least one tumor episode diagnosed during the follow-up period, whereas disease free patients had a history of TCC but no active disease during this period, according to the clinical data. The study period went from January 1991 to February 1994, during which time urine specimens were collected and clinical procedures were performed. All data were collected from medical records; hence, the investigators had no direct patient contact. Clinical disease status was determined from the results of biopsy and cystoscopy as reported in medical records. These clinical data were used as the "gold standard" against which laboratory data were compared. Laboratory data were recorded for ten time periods prior to the end of the study period. The length of these time periods, defined by the interval between the laboratory test and clinical observation, generally was chosen as a 3-month period to be consistent with the expected length of time between follow-up visits for a patient experiencing a recent disease episode. Inclusion Criteria The inclusion criteria were: 1) a positive biopsy within the study period (case) or a history of TCC prior to the study period but with no evidence of disease during the study period (control); 2) at least three DNA image analyses on urine specimens simultaneous with visual cytology; 3) the patient's urologist began using the DNA image analysis before 1991; 4) the patient's urologist responded to the first mailing; and 5) the patient underwent at least one cystoscopy or biopsy during the designated study period. Exclusion Criteria Exclusion criteria were: 1) bladder carcinoma of a histologic type other than TCC; 2) no cystoscopy or biopsy performed during the study period; and/or 3) sufficiently poor data quality so that the disease status could not be determined. Cystectomy was not a criterion for exclusion of patients, who were still at risk for developing TCC in the ureters or the renal pelvis of the kidneys. Patients with TCC in areas of the genitourinary tract other than the bladder were not excluded because DNA image analysis measured TCC cells exfoliated in the urine originating from any site in the urinary tract. Definition of Endpoints Biopsy results were considered the gold standard for determining the disease status for bladder carcinoma, except when no biopsy was performed. When no biopsy was performed, a negative cystoscopy was accepted as indicating that no tumor was present. A positive cystoscopy and subsequent treatment for tumor recurrence were accepted as evidence of recurrent disease. Study Period The start of the study period began at the patient's "reference date," which was the date of the patient's first laboratory test after January 1, 1991. For participants with recurrent disease, the follow-up period extended until the first date of diagnosis of tumor recurrence after the reference date. For disease free participants, the endpoint was defined by the last follow-up visit to the physician's office prior to February 28, 1994. Clinical Data Collection Methods Data were collected by two methods. First, for 60 patients in Chicago, New York, and Washington, DC, data were abstracted by one of us (A.R.) from medical charts in 5 doctors' offices. The selection of these offices was based on location (proximity to the diagnostic laboratory or a large number of eligible patients in one office) and willingness of the urologist to allow data abstraction of his or her charts. Data on the other 115 patients, under the care of 45 urologists, were collected by means of the second mailing. Doctors were asked to complete a data collection form on each of their patients included in the study. The data collection form was individualized for each patient, based on the reference date. It requested information regarding the dates and results of biopsies and cystoscopies performed during the study period, examined prospectively from the reference date. Pathology data included the date of the first biopsy on or after the reference date, the histologic type, grade, and stage of the tumor, and the pattern of growth (i.e., papillary, carcinoma in situ [CIS], or solid). Cystoscopic information included the date of each examination during the study period, whether or not a tumor was visualized, the stage of the tumor, and the pattern of growth. Telephone calls were made to the physicians' offices for clarification when data were inconsistent internally. Those patients whose disease status or clinical endpoints were unclear and could not be validated were excluded from the study. To maintain the confidentiality of patients and physicians, no personal identifiers were entered into the data set for analysis. Each patient was given a study code number. Clinical Procedures Performed during Follow-Up During the follow-up period, every patient underwent at least 1 clinical procedure; 36% underwent at least 1 biopsy. Information was collected on all 847 cystoscopies reported to have been performed during the study period, 85 of which (10%) were positive. "Suspicious" cystoscopic findings were interpreted as negative. The number of cystoscopies per patient ranged from 1-10 with an average of 4.8. Biopsies were recorded for 63 patients, 45 of which (71%) were positive. Only the first biopsy performed during the study period was recorded for purposes of this analysis because it served as an individual endpoint for follow-up of the patient. Laboratory Procedures Laboratory results, including laboratory diagnosis and aneuploidy score (AS), were obtained from the mainframe data base at the diagnostic laboratory. The term "laboratory diagnosis" or "laboratory test" is used to refer to the combination of visual cytology and image analysis results that were obtained as a single score from the database. Data for image analysis alone are presented; however, data for visual cytology alone are not. For the DNA analysis, urine specimens (voided urine, catheterized urine, or bladder washings) had been collected, fixed, and prepared. Most urine specimens were voided urine but catheterized urine and bladder washings were accepted when invasive procedures were performed for reasons other than this test. The standard procedures of the Feulgen staining method were used to stain the cells.29, 30 Visual cytology Visual cytologic evaluations initially were made by cytotechnologists, who recorded a preliminary diagnosis using standard codes commonly used to report cytologic findings. The criteria used for assessment were standard, accepted criteria, including nuclear to cytoplasmic ratio, intensity of staining, texture of chromatin, definition of boundaries, and degree of differentiation.5, 10, 31 Image cytometry The Leitz Autoplan image analyzer was used (Leitz, Wetzlar, Germany). The same slides used for cytology were stained for DNA image analysis; the same group of cells was examined using each method. A histogram was produced showing the ploidy of the cells, and an AS quantified the degree of abnormality of the DNA. This score was calculated according to the method of Bocking et al., in which the amount of deviation from the normal 2c state was measured.32 This score then was converted from a geometric scale to a logistic scale to create an AS range from 0-100. A score of ≥ 25 was considered abnormal. Laboratory diagnosis A laboratory diagnosis was given on a scale of one to five that also was used for the visual cytologic evaluations: 1 = negative, 2 = reactive, 3 = atypical, 4 = suspicious, and 5 = positive. Initial diagnosis was made by a cytotechnologist. A final diagnosis was made by a pathologist who reviewed the 1) slides, 2) cytotechnologist's visual cytology report, 3) histogram, and 4) AS. The quality of the specimen, such as the number of cells and morphologic effects of treatment, were noted to improve the validity of diagnosis. Data Analysis The goal of the analysis was to determine the ability of the DNA image analysis, in combination with visual cytology, to predict disease status at different points in time prior to clinical evidence of disease. Data were entered using EpiInfo Version 5 33 and were analyzed using the Statistical Analysis System Version 6.34 The strategy of analysis involved classifying all patients as either recurrent or disease free, based on the clinical evidence. The diagnostic tests then were classified into time intervals, based on the number of days by which the laboratory test preceded the clinical endpoint. The taxonomy for the time intervals was: < 2 days (simultaneous); 2-30 days; 31-60 days; 61-90 days; 91-80 days; 7-9 months; 10-12 months; 13-15 months; 16-18 months; and > 18 months. Sensitivity and specificity Two-by-two tables of clinical disease status by diagnostic laboratory test results were generated for each time interval. Sensitivity, specificity, false-positive and false-negative error rates, and positive and negative predictive values were determined for each table.35 Sensitivity and specificity are reported in the current study. The true disease status was either positive (cases with recurrent disease) or negative (disease free controls). Disease status was used as the standard against which to compare two sets of laboratory results: the laboratory diagnosis and the AS. The 2 x 2 tables were repeated for different cutoff points, corresponding to different levels of rigor in assigning a diagnosis. A dichotomous variable was created from the laboratory diagnosis by cutting the ordinal variable at different points: 1 (negative) versus 2-5; 1-2 versus 3-5; 1-3 versus 4-5; and 1-4 versus 5 (positive). The AS alone was made into a dichotomous variable by considering a score of 15+ positive, then 20+ = positive, then 25+ = positive, and finally 30+ = positive. Receiver operating characteristic curves Receiver operating characteristic (ROC) curves were used to assess the strength of the diagnostic test as a predictor for bladder carcinoma recurrence in all patients studied, and to determine the interval before diagnosis for which the screening test was useful. ROC curves also were used to compare the relative ability of the laboratory diagnosis (including both DNA image analysis and visual cytology) and the AS score, which is determined from only the DNA image analysis, to predict the true status. RESULTS Mailing Results A total of 102 of 195 questionnaires from the first mailing were returned by the urologists (52%). Overall responses regarding 423 patients showed that 359 (85%) had a history of TCC. Of 403 responses to the second question, 198 patients were reported to have had active TCC (49%), defined as positive biopsy or positive cystoscopy when no biopsy was performed. Forty-eight of 113 second questionnaires were completed and returned by mail (42%). These contained data on 147 patients, compared with the 418 patients requested (35%). Thirty-two patients were excluded (21.8%), 23 for lack of clinical follow-up and 9 due to inadequate data. The remaining 115 patients whose data were collected by questionnaire were included in the study. Of the 60 patients whose data were collected by chart abstraction, none were excluded. This left 175 patients with adequate data (37% of the total 478 patients for whom data were sought). Descriptive Statistics Demographics Of 175 patients, 126 (72%) were male and 49 (28%) were female. Their ages ranged from 32-94 years, with a mean age of 69.8 years (standard deviation = 11.3 years); there was no statistically significant difference between the ages of men and women. The study subjects were patients of 50 different private physicians. One physician had 46 patients included in the study, and each of the other physicians had between 1 and 6 patients included. The study patients lived in all parts of the U. S.: 34% were from the Central region, 19% from the Mid-Atlantic states, 10% from the Northeast, 18% from the Pacific region, 9% from the Southeast, and 10% from the Southwest. Disease characteristics Of the 175 study subjects, 115 remained disease free, whereas 60 patients had a positive diagnosis of TCC during the study period. Of the 60 clinically diseased patients, 45 (75%) were confirmed by biopsy, and 15 (25%) were diagnosed by cystoscopy alone. Six of the 15 patients diagnosed by cystoscopy alone had at least 1 negative biopsy during the follow-up time, prior to the diagnosis. Of the 115 patients considered clinically negative, 12 were biopsied with negative results, and 103 were diagnosed as negative by cystoscopy. Among the 60 patients with recurrent disease, 50 (83%) had superficial tumors, 5 (8%) had smooth muscle invasion, and the pattern of invasion was unclear for 5. Of the 50 superficial tumors, 30 were specified further by tumor classification as determined by biopsy: 8 (27%) were Tis (CIS), 19 (63%) were Ta (no invasion of the lamina propria), and 3 (10%) were T1 (invasion of the lamina propria but not the muscularis) (TNM classification system).36 Of the 45 cases determined by biopsy, 13 (29%) were Grade I (well differentiated), 15 (33%) were Grade II (moderately well differentiated), 11 (24%) were Grade III or IV (poorly differentiated), and the grade was not reported for 6 (13%). Among the 60 patients with recurrent disease, 37 (62%) exhibited papillary growth patterns alone, 2 (3%) had solid tumors alone, 6 (10%) had CIS alone, 3 (5%) had CIS and papillary growth patterns, 2 (3%) had solid tumors and CIS, and 10 (17%) were not categorized due to incomplete data. Of the tumors for the 15 cases determined by cystoscopy alone, 8 were reported as papillary, 1 was reported as CIS, and 6 were not specified. Chart abstraction versus mailing Of the 175 study subjects, data were collected by chart abstraction for 60 and by mailing for 115. The proportion of positive cases in each group is equivalent: 35% in the chart abstraction group compared with 34% in the mailing group; however, biopsy confirmation of disease was more likely in the chart review group (86% vs. 69%). Both groups had a high proportion of superficial tumors; however, missing data make comparison of percentages unreliable. Effect of Time and Stringency of Criteria for a Positive Test on Sensitivity and Specificity The sensitivity and specificity of the laboratory diagnosis, resulting from the combined visual cytology and image analysis test, are shown at various cutpoints as a function of time in Table 1 and Figures 1 (6K) and 2 (5K). Time intervals, shown on the x axis, indicate the time from the laboratory test date to the date of diagnosis. The graphed lines show sensitivity and specificity over time at four different levels of stringency for defining a positive laboratory diagnosis. Sensitivity and specificity of the AS at various cutpoints are shown in Table 2 and Figures 3 (5K) and 4 (5K). As hypothesized, as the time from the test to the time of clinical diagnosis increased (up to approximately 9 months), the sensitivity of the laboratory diagnosis decreased and, generally, specificity increased. Figure 1Open in figure viewerPowerPoint Sensitivity of laboratory diagnosis for transitional cell carcinoma of the bladder at four cutoff points as a function of time. Laboratory diagnosis (1 = negative, 2 = reactive, 3 = atypical, 4 = suspicious, and 5 = positive) represents the combination of image analysis and visual cytology. The time interval represents the time between the laboratory test and clinical observation. Figure 2Open in figure viewerPowerPoint Specificity of laboratory diagnosis for transitional cell carcinoma of the bladder at four cutoff points as a function of time. Laboratory diagnosis (1 = negative, 2 = reactive, 3 = atypical, 4 = suspicious, and 5 = positive) represents the combination of image analysis and visual cytology. The time interval represents the time between the laboratory test and clinical observation. Figure 3Open in figure viewerPowerPoint Sensitivity of the aneuploidy score for transitional cell carcinoma at four cutoff points as a function of time. The aneuploidy score represents a quantitative measure of image analysis ranging from 0-100. The time interval represents the time between image analysis and clinical observation. Figure 4Open in figure viewerPowerPoint Specificity of the aneuploidy score for transitional cell carcinoma at four cutoff points as a function of time. The aneuploidy score represents a quantitative measure of image analysis ranging from 0-100. The time interval represents the time between image analysis and clinical observation. Table 1. Sensitivity and Specificity of Laboratory Diagnosis (DNA Image Analysis plus Visual Cytology) at Different Time Intervals and Cutoff Points for all Patients (n = 175) Cutoff point 1 vs. 2+ 1-2 vs. 3+ 1-3 vs. 4+ 1-4 vs. 5 Time interval Sens Spec Sens Spec Sens Spec Sens Spec 0-1 day 80.4 41.1 41.1 78.2 23.2 91.1 7.1 97.3 2-30 days 68.5 38.4 38.9 81.4 25.9 91.9 9.3 96.0 31-60 days 59.6 41.5 34.0 77.8 21.3 93.9 8.5 96.0 61-90 days 56.4 43.6 28.2 85.5 20.5 92.7 7.7 96.3 91-180 days 33.9 75.9 19.6 90.0 7.4 94.6 7.1 95.5 7-9 mos 25.9 75.7 11.1 86.6 7.4 88.9 7.4 88.9 Sens: sensitivity; Spec: specificity. Table 2. Sensitivity and Specificity of Aneuploidy Score at Different Time Intervals and Cutoff Points for all Patients (n = 175) Cutoff point < 14 vs. 15+ < 19 vs. 20+ < 24 vs. 25+ < 29 vs. 30+ Time interval Sens Spec Sens Spec Sens Spec Sens Spec 0-1 days 30.9 82.9 38.2 80.2 30.9 82.6 29.1 86.5 2-30 days 37.7 83.7 41.5 77.6 37.7 83.3 26.4 86.7 31-60 days 27.7 87.7 29.8 79.0 27.7 82.5 25.5 87.7 61-90 days 21.1 86.8 23.7 86.8 21.1 86.8 18.4 86.8 91-180 days 46.3 64.2 46.3 60.4 46.3 64.2 43.9 64.2 7-9 mos 46.9 57.1 46.9 54.8 46.9 57.1 46.9 57.2 Sens: sensitivity; Spec: specificity. In addition, as generally was expected, sensitivity decreased and specificity increased with increasing stringency in the criteria for a positive laboratory diagnosis. Sensitivity and specificity also were affected by changes in the cutoff point for the AS, although by much less than by changes in the cutoff point for the combined laboratory diagnosis. When the laboratory diagnosis was made at essentially the same point in time as the clinical diagnosis (interval 1), the striking effects of stringency of the laboratory test criteria on sensitivity are evident in Table 1 and Figure 1 (6K). The sensitivity of the laboratory diagnosis decreased from 80.4% (when the cutoff point for the test = 1 vs. 2-5) to 7.1% (when using a cutoff point of 1-4 vs. 5). Likewise, the specificity increased from 41% at a cutoff point of 1 versus 2-5 to 97% at a cutoff point of 1-4 versus 5. In addition, keeping the cutoff point for the laboratory test at 1 versus 2-5, the sensitivity of the laboratory test fell from 80.4% when the test and clinical diagnosis were made at essentially the same time to 25.9% when the test was done 7-9 months before the diagnosis of recurrence was made. There was little effect of a longer time interval on sensitivity when the test cutoff point was set at 1-4 versus 5, because the sensitivity was very low to begin with. ROC Curve An ROC curve was developed for the combined laboratory diagnosis to determine which cutoff point for the laboratory reports showed the best operating characteristics (Fig. 5 (5K)). An ideal test would show an ROC curve that rose along the left-hand y axis and touched the upper left corner, at which point the sensitivity would be 1.0 (100%) and the false-positive error rate would be 0.0 (so that the specificity also would be 100%). Figure 5Open in figure viewerPowerPoint Receiver operating characteristics (ROC) curve for the laboratory diagnosis (DX) for transitional cell carcinoma of the bladder (TCC) at the simultaneous time period (0-1 day). Laboratory diagnosis represents the combination of image analysis and visual cytology. The area under the curve represents the strength of the test as a predictor of TCC recurrence. The combined test showed a disappointing ROC curve. Its operating characteristics were not much better than the line of no benefit (a straight line from the lower left corner to the upper right corner). The ROC curve in Figure 5 (5K) plots the four possible cutoff points at the simultaneous time interval. The 1 versus 2-5 cutoff point at the simultaneous time interval showed a higher test sensitivity than did any other combination of characteristics. ROC curves were developed for all the other cutoff points and time intervals, none of which were as good as the one shown in Figure 5 (5K). DISCUSSION Image analysis in combination with visual cytology is supported for use in monitoring patients for recurrence of bladder carcinoma and response to treatment. The time element in this study showed that DNA image analysis plus visual cytology may be indicative of malignant changes up to 3 months prior to discovery by clinical follow-up. The sensitivity of the test remained > 50% up to 3 months prior to the time of clinical diagnosis. The laboratory test can be used with noninvasively obtained urine samples as a screening test for patients with a history of TCC to determine when unplanned cystoscopies may be needed; it is not intended a