Quantitative light microscopy in urological oncology

Cultivating Clinical Clarity through Computer Vision: A Current Perspective on Whole Slide Imaging and Artificial Intelligence

Diagnostic devices, methodological approaches, and traditional constructs of clinical pathology practice, cultivated throughout centuries, have transformed radically in the wake of explosive technological growth and other, e.g., environmental, catalysts of change. Ushered into the fray of modern laboratory medicine are digital imaging devices and machine-learning (ML) software fashioned to mitigate challenges, e.g., practitioner shortage while preparing clinicians for emerging interconnectivity of environments and diagnostic information in the era of big data. As computer vision shapes new constructs for the modern world and intertwines with clinical medicine, cultivating clarity of our new terrain through examining the trajectory and current scope of computational pathology and its pertinence to clinical practice is vital. Through review of numerous studies, we find developmental efforts for ML migrating from research to standardized clinical frameworks while overcoming obstacles that have formerly curtailed adoption of these tools, e.g., generalizability, data availability, and user-friendly accessibility. Groundbreaking validatory efforts have facilitated the clinical deployment of ML tools demonstrating the capacity to effectively aid in distinguishing tumor subtype and grade, classify early vs. advanced cancer stages, and assist in quality control and primary diagnosis applications. Case studies have demonstrated the benefits of streamlined, digitized workflows for practitioners alleviated by decreased burdens.

Evaluation of ICSI-Selected Epididymal Sperm Samples of Obstructive Azoospermic Males by the CKIA System

The objective of this study was to evaluate nuclear normality in intracytoplasmic sperm injection (ICSI)-selected epididymal sperm from obstructive azoospermic (OA) patients. We evaluated whether the selection criteria used in routine ICSI (morphology and motility at a magnification of 400x) is adequate for selecting "normal" sperm from epididymal samples. Surgically retrieved spermatozoa from the caput epididymis of 15 OA patients and ejaculated sperm samples from 9 normospermic donors were evaluated with a DNA-specific stain (Feulgen) and in combination with the computerized karyometric image analysis (CKIA) system. Original (unselected) samples and ICSI-selected sperm were compared in donor and patient samples. In the original fraction, a larger variation in almost all measured parameters was found in epididymal sperm than in ejaculated sperm. After sperm selection, the morphometry was comparable between epididymal and ejaculated sperm. However, for those parameters related to the DNA stainability and chromatin texture (nuclear condensation), significant differences between patients and donors were observed. This result suggests that the size and form of the sperm do not necessarily hold similar internal structures. Thus, the frequency of "normal" sperm significantly increased after ICSI selection, but the improvement was more marked in donor than in OA patients' samples. In conclusion, at least a twofold increase in the number of normal spermatozoa was achieved after ICSI selection. The heterogeneity in the stainability and chromatin condensation of epididymal samples from OA patients indicates that some of the selected spermatozoa have a hypocondensed or hypercondensed chromatin. Even in the best of donor cases, no more than 55% of the selected sperm scored normal with CKIA, indicating that the present routine ICSI selection criteria are not sufficient for selecting normal condensed nuclei.

Cell-nuclear data reduction and prognostic model selection in bladder tumor recurrence

OBJECTIVE

The paper aims at improving the prediction of superficial bladder recurrence. To this end, feedforward neural networks (FNNs) and a feature selection method based on unsupervised clustering, were employed.

MATERIAL AND METHODS

A retrospective prognostic study of 127 patients diagnosed with superficial urinary bladder cancer was performed. Images from biopsies were digitized and cell nuclei features were extracted. To design FNN classifiers, different training methods and architectures were investigated. The unsupervised k-windows (UKW) and the fuzzy c-means clustering algorithms were applied on the feature set to identify the most informative feature subsets.

RESULTS

UKW managed to reduce the dimensionality of the feature space significantly, and yielded prediction rates 87.95% and 91.41%, for non-recurrent and recurrent cases, respectively. The prediction rates achieved with the reduced feature set were marginally lower compared to the ones attained with the complete feature set. The training algorithm that exhibited the best performance in all cases was the adaptive on-line backpropagation algorithm.

CONCLUSIONS

FNNs can contribute to the accurate prognosis of bladder cancer recurrence. The proposed feature selection method can remove redundant information without a significant loss in predictive accuracy, and thereby render the prognostic model less complex, more robust, and hence suitable for clinical use.

Advanced soft computing diagnosis method for tumour grading

OBJECTIVE

To develop an advanced diagnostic method for urinary bladder tumour grading. A novel soft computing modelling methodology based on the augmentation of fuzzy cognitive maps (FCMs) with the unsupervised active Hebbian learning (AHL) algorithm is applied.

MATERIAL AND METHODS

One hundred and twenty-eight cases of urinary bladder cancer were retrieved from the archives of the Department of Histopathology, University Hospital of Patras, Greece. All tumours had been characterized according to the classical World Health Organization (WHO) grading system. To design the FCM model for tumour grading, three experts histopathologists defined the main histopathological features (concepts) and their impact on grade characterization. The resulted FCM model consisted of nine concepts. Eight concepts represented the main histopathological features for tumour grading. The ninth concept represented the tumour grade. To increase the classification ability of the FCM model, the AHL algorithm was applied to adjust the weights of the FCM.

RESULTS

The proposed FCM grading model achieved a classification accuracy of 72.5%, 74.42% and 95.55% for tumours of grades I, II and III, respectively.

CONCLUSIONS

An advanced computerized method to support tumour grade diagnosis decision was proposed and developed. The novelty of the method is based on employing the soft computing method of FCMs to represent specialized knowledge on histopathology and on augmenting FCMs ability using an unsupervised learning algorithm, the AHL. The proposed method performs with reasonably high accuracy compared to other existing methods and at the same time meets the physicians' requirements for transparency and explicability.

Gleason grading and prognostic factors in carcinoma of the prostate

Gleason grade of adenocarcinoma of the prostate is an established prognostic indicator that has stood the test of time. The Gleason grading method was devised in the 1960s and 1970s by Dr Donald F Gleason and members of the Veterans Administration Cooperative Urological Research Group. This grading system is based entirely on the histologic pattern of arrangement of carcinoma cells in H&E-stained sections. Five basic grade patterns are used to generate a histologic score, which can range from 2 to 10. These patterns are illustrated in a standard drawing that can be employed as a guide for recognition of the specific Gleason grades. Increasing Gleason grade is directly related to a number of histopathologic end points, including tumor size, margin status, and pathologic stage. Indeed, models have been developed that allow for pretreatment prediction of pathologic stage based upon needle biopsy Gleason grade, total serum prostate-specific antigen level, and clinical stage. Gleason grade has been linked to a number of clinical end points, including clinical stage, progression to metastatic disease, and survival. Gleason grade is often incorporated into nomograms used to predict response to a specific therapy, such as radiotherapy or surgery. Needle biopsy Gleason grade is routinely used to plan patient management and is also often one of the criteria for eligibility for clinical trials testing new therapies. Gleason grade should be routinely reported for adenocarcinoma of the prostate in all types of tissue samples. Experimental approaches that could be of importance in the future include determination of percentage of high-grade Gleason pattern 4 or 5, and utilization of markers discovered by gene expression profiling or by genetic testing for DNA abnormalities. Such markers would be of prognostic usefulness if they provided added value beyond the established indicators of Gleason grade, serum prostate-specific antigen, and stage. Currently, established prognostic factors for prostatic carcinoma recommended for routine reporting are TNM stage, surgical margin status, serum prostate-specific antigen, and Gleason grade.

Bcl-2 correlates with tumor ploidy and nuclear morphology in early stage prostate carcinoma. A fine needle aspiration biopsy study

We evaluated the nuclear morphology, ploidy, bcl-2 expression and in situ apoptosis in sections of fine-needle aspiration (FNA) biopsy specimens of thirty-one randomly selected Stage B prostate carcinomas. Sections of paraffin-embedded pelleted cells obtained from FNA biopsy specimens were studied. Nuclear grade was determined according to the WHO system. Nuclear morphometry and DNA ploidy were carried out using an automated image analyzer. We used immunostaining and the TUNEL method to evaluate bcl-2 expression and in situ apoptosis. The median nuclear area increased with increasing nuclear grade. Ploidy analysis showed that 54.8% of tumors were diploid, 3.2% tetraploid and 41.9% aneuploid. Bcl-2 overexpression was found in 10 of 31 tumors. There was a significant positive correlation between bcl-2 expression and nuclear area (r(s): 0.45 p < 0.01). Nine of ten bcl-2-positive tumors had a nuclear area larger than the median of the series, and 70% of bcl-2-positive tumors were of the aneuploid type. The apoptotic index had a negative correlation with nuclear area, and the lowest indexes were found in aneuploid tumors. Bcl-2 expression showed a highly significant association with both parameters of high aggressiveness: nuclear size and aneuploidy. The combined evaluation of nuclear morphology, ploidy and cell survival parameters might better identify patients with poor prognosis among early stage prostate carcinomas diagnosed by FNA biopsies.

Neural network-based digitized cell image diagnosis of bladder wash cytology

In this pilot study, we tested whether it is possible to apply neural network-based diagnostics on bladder washings to detect urothelial cell carcinoma of the bladder. Eighty-five bladder wash (BW) samples were chosen at random from our own database. Cystoscopy, histology, and follow-up data concerning tumor recurrence were available. Each slide was scanned by the neural network-based digitized cell image system. The neural network-based diagnosis (NNBD) was based on 128 digitized cell images provided by the system. The light microscopic diagnosis (LMD) was rendered by an experienced cytopathologist using the same terminology, i.e., negative, low-grade tumor, and high-grade tumor. Finally, an automatic QUANTICYT analysis was performed on the same material, with as classification low, intermediate, and high risk. The sensitivity for diagnosing a histologically confirmed tumor was for NNBD 92%, for LMD 50%, and for QUANTICYT 69%. For the three methods, receiver operating characteristic (ROC) curves were made for the thresholds low grade/intermediate risk and high grade/high risk. For the prediction of a positive cystoscopy, the highest area under the curve (AUC) was found for NNBD, being 0.71. The AUC for LMD was 0.58. QUANTICYT analysis had the highest AUC value (0.62) for predicting tumor recurrence after a negative cystoscopy, with a lower value for NNBD (0.50). These findings indicate that neural network-based diagnosis of bladder washing samples is highly promising.

Presence of carcinoma in situ and high 2C-deviation index are the best predictors of invasive transitional cell carcinoma of the bladder in patients with high-risk Quanticyt

OBJECTIVES

Karyometric analysis (Quanticyt) has proved of value as a cytologic marker for bladder cancer. This study was conducted to identify diagnostic and prognostic factors in a high-risk Quanticyt population to predict the prognosis of transitional cell carcinoma (TCC) of the bladder.

METHODS

Quanticyt is a karyometric system for quantitative bladder wash cytologic findings based on two nuclear features: the 2c-deviation index (2cDI) and the mean of nuclear shape. Samples are scored as low, intermediate, or high risk. Before 1995, 109 patients with high-risk quantitative bladder wash cytologic findings were identified at our clinic. Four patients with previous invasive tumors were excluded.

RESULTS

Histologically proven malignancy was found in 54 of 105 patients at first high-risk quantitative bladder wash cytologic findings. Invasive TCC was found in 16 patients, and another 10 patients had progression during a median follow-up of 3.7 years. In univariate analysis, the presence of carcinoma in situ (CIS), highest tumor grade, 2cDI, and highest tumor stage were significant predictors of progression. The presence of CIS proved to be the only predictor of progression in the multivariate analysis. A 2cDI of 2.00 c(2) or higher was a significant predictor of CIS, invasive TCC, and progression. At follow-up analysis after negative cystoscopy, 2cDI showed a tendency toward predicting progression.

CONCLUSIONS

These data confirm earlier findings that CIS is an important marker of progression. 2cDI as assessed by quantitative cytology is a practical tool to predict CIS, invasive TCC, and subsequent progression. A 2cDI of 2. 00 c(2) can be used to further stratify high-risk quantitative bladder wash cytologic findings.

Use of Lewis X antigen and deoxyribonucleic acid image cytometry to increase sensitivity of urinary cytology in transitional cell carcinoma of the bladder

PURPOSE

To improve sensitivity and specificity of urinary cytology for bladder cancer cell detection we used deoxyribonucleic acid (DNA) image cytometry and the monoclonal anti-Lewis X antibody P12.

MATERIALS AND METHODS

Spontaneously voided urine and additional barbotage bladder washings of 25 patients with transitional cell carcinomas and 25 patients with benign diseases of the bladder were analyzed by conventional cytology, immunocytology and DNA image cytometry. The DNA content was determined in specimens stained with Feulgen according to the European Society for Analytical Cellular Pathology consensus report on standardization of DNA image cytometry. For the immunocytological examination we used the avidin-biotin-complex immunoperoxidase method with the monoclonal antibody P12 directed against the Lewis X determinant.

RESULTS

The cytological examination revealed a sensitivity of 72% and a specificity of 100%. By using DNA image cytometry with Kolmogoroff-Smirnow test sensitivity increased up to 84% with a specificity of 100%. The immunocytological examination with Lewis X showed a sensitivity of 84% and a specificity of 80%.

CONCLUSIONS

Combining cytology and DNA cytometry the overall sensitivity increased to 92% and with the additional application of immunocytology for Lewis X antigen sensitivity it increased to 96% but was accompanied by a decrease in specificity to 80%. DNA image cytometry and Lewis X antigen detection are not suitable for screening but suspicious urothelial cells in urinary cytology specimens can be evaluated specifically by DNA measurements. In the future it needs to be clarified whether DNA image cytometry in combination with Lewis X antigen detection can help to signal relapse and progression of transitional cell carcinoma of the bladder.

Combining QUANTICYT karyometric analysis with architectural confocal-aided cytology to prognosticate superficial bladder cancer

For this study, bladder washings of 923 patients with a follow-up for superficial bladder cancer were used. For the prognostication of each case (follow-up period up to 20 mo), two parameters were used: tumor recurrence and the event of (progressing) invasive growth. Each sample was classified as negative, low grade or high grade, based on confocal-aided architectural cytology. The feature architecture was based on the images of the epithelial fragments as seen in the Feulgen stain, which also could be visualized in confocal microscopy. The same three classes were also used for grading the samples according to the QUANTICYT system, in which DNA and morphometry of nuclei were automatically assessed. The 532 cases graded as QUANTICYT high had a worse prognosis than the 391 cases with QUANTICYT low. In addition, architectural cytology contributed significantly for the prognostication of cases classified as QUANTICYT high, the 228 nonconcording cases faring better than the 304 concording ones. For QUANTICYT low, the 156 nonconcording cases had also a lower recurrence rate than the 235 concording ones, but this difference was only significant in patients followed for up to 8 mo. These findings indicate that QUANTICYT grading supplemented by architectural cytology supplies the clinician with valuable prognostic information which can be used to limit cystoscopies.

Quanticyt: karyometric analysis of bladder washing for patients with superficial bladder cancer

OBJECTIVES

Quantitative cytology by image-analysis techniques enables objective interpretation of nuclear features in light microscopic images. QUANTICYT, a quantitative karyometric cytology system, was used in the follow-up of patients with superficial bladder cancer.

METHODS

From 1992 to 1995, 4137 samples from 1412 patients were obtained. At 1-year follow-up after the initial bladder washing, a tumor recurrence rate of 21% was found. In this period, tumor progression to invasive disease occurred in 1.6% of patients. Scoring of tumor by the QUANTICYT system was based on two nuclear features: the 2c deviation index and the mean of a nuclear shape feature: MPASS.

RESULTS

The method was found to be reproducible and superior to visual cytologic interpretation. QUANTICYT analysis of the bladder washings resulted in a score of low, intermediate, and high risk. In a multivariate analysis, highest grade of earlier tumor and QUANTICYT risk score were the best predictors of tumor recurrence and progression. For the easy application of QUANTICYT analysis in daily routine, a report form that included patient history and DNA histogram was developed.

CONCLUSIONS

QUANTICYT karyometric analysis of bladder-wash material proved a useful, clinically applicable grading tool in the follow-up of patients with superficial bladder cancer, with sufficient power to be used in decision-making in the individual patient.

The use of the digital cell image analysis of Feulgen-stained nuclei to detect apoptosis

Cell death is an essential event in the functioning of multicellular organisms. It plays a role opposite to that of mitosis in the regulation of cell populations. In the present work, we describe an original methodology which permits the easy detection and count of apoptotic cells in a given tissue. This methodology is based on the digital cell image analysis of Feulgen-stained nuclei, which also permits the calculation of the proliferation index, i.e. the percentage of cells in the S phase of the cell cycle. This percentage of cells in the S phase is strongly related to the mitotic index. Our methodology, which involves the multivariate analysis of 14 morphonuclear parameters computed by means of the digitized cell image analysis of Feulgen-stained nuclei, was applied here to a well-known biological apoptosis model, namely glucocorticoid-treated rat thymocytes. The parameters that permitted the detection of apoptotic cells were the integrated optical density, a parameter that describes the nuclear DNA content, and the run length percentage and long run length parameters which are related to the pattern of chromatin condensation. This determination can be carried out on a relatively small number of cells.

Basic research

Basic research is currently investigating the molecular cascade associated with bladder cancer development. Many new findings are potential leads towards the improvement of the diagnosis and prognosis of this disease. Special care, however, should be taken in the design of protocols for clinical evaluation of the value of these markers. Some initial guidelines have been put forward in this report.

The use of digital image analysis of chromatin texture in Feulgen-stained nuclei to predict recurrence of low grade superficial transitional cell carcinoma of the bladder

BACKGROUND

Identifying a marker enabling prediction of recurrence in the group of superficial transitional cell carcinomas (sTCCs) of the bladder remains an important challenge today. This report quantitatively describes chromatin patterns with respect to such sTCC recurrence.

MATERIALS AND METHODS

Twenty-nine patients with sTCCs who did not exhibit tumor recurrence within a minimum of 24 months were compared with 21 patients with sTCCs who exhibited tumor recurrence two or three times in a 24-month period, for a total of 74 sTCCs. Quantitative chromatin pattern description was performed by the digital cell image analyses of Feulgen-stained nuclei. Six morphonuclear parameters were thus described and subsequently used to determine a score, allowing biological behavior of sTCCs to be described, i.e., recurrence versus non-recurrence in one calculation step. DNA ploidy level was also determined in each sTCC by assessing its DNA histogram type.

RESULTS

Of 32 patients with Grade 1 pathologically classified pTa/pT1 tumors, DNA ploidy level determination permitted correct prediction of tumor nonrecurrence or recurrence of 13 (41%), whereas determination of the score values enabled prediction of nonrecurrence or recurrence of 25 (78%). Combining DNA ploidy level data and the score values enabled recurrence or nonrecurrence to be predicted for 29/32 of the patients (91%).

CONCLUSIONS

The quantitative description of chromatin patterns by digital cell image analysis of Feulgen-stained nuclei can provide helpful information, in addition to DNA ploidy level determination, in predicting tumor recurrence of low grade superficial transitional cell carcinomas of the bladder.

Preoperative histological and cytological diagnosis and DNA ploidy assessment of localized prostatic cancer. A comparison between findings in biopsies and total prostatectomy specimens

The preoperative biopsy specimens of 49 men who underwent radical prostatectomy were compared with serially step-sectioned operative specimens, in order to evaluate diagnostic accuracy as regards the presence of cancer, its grade and its DNA ploidy. Both the FNAB and TRUS-GCB undergraded the cancer by one WHO grade in about 35-40% of cases. In cases where both biopsy types were available, the degree of undergrading was the same. Determination of DNA ploidy by the single-cell technique from FNAB had a sensitivity for detecting non-diploid DNA patterns of 59% whereas flow cytometric measurement of core biopsies had a sensitivity of 44.4% as regards non-diploid DNA, when compared with operative specimens. A comparison of ploidy in core biopsies versus ploidy in fine-needle aspirates revealed that more non-diploid DNA patterns were diagnosed in the fine-needle aspirates. These aneuploid patterns were not all confirmed by flow cytometric evaluation of the operative specimens, in which, however, more aneuploid patterns were diagnosed compared with the single-cell technique from FNAB. We conclude that the standard technique with 1-3 fine-needle aspirates or 1-3 TRUS-GCB has a pronounced tendency by both biopsy methods to underestimate the tumour grade. DNA ploidy analysis by FNAB and the single-cell cytometric method reveals aneuploid cell lines not found in the flow cytometric evaluation. In order to determine whether this reflects a methodological problem or a true discrimination between the ability of the two methods to find non-diploid cell lines further studies are needed. However, for a more correct preoperative assessment of tumour grade and DNA ploidy, more extensive sampling is required.

Estrogen treatment postpones the castration-induced dedifferentiation of Dunning R3327-PAP prostatic adenocarcinoma

Male Copenhagen x Fisher F1 rats, transplanted with the androgen-sensitive Dunning R3327 PAP rat prostatic adenocarcinoma, were castrated when tumor volumes were approximately 1300 mm3. The rats were thereafter followed with measurements of tumor volume. Castration stopped tumor growth, but some of the tumors started to regrow after 7-36 weeks. These tumors relapsing from castration treatment were now considered to be androgen-insensitive. In this study, we defined relapse as the time when the tumor volume had increased to 200% of the volume at the time for castration. At this time, the rats were treated either with estradiol-17 beta (E2, 50 micrograms s.c. daily) or vehicle for 8 weeks. After this period, tumor morphology was examined. The tumors in the vehicle-treated group were heterogeneous, and both highly and more dedifferentiated parts were present. The tumor growth rate was correlated to the epithelial cell nuclear size and its variance, and to the mitotic index. In the E2-treated group, tumor growth rate was retarded throughout the treatment period, and dedifferentiated tumor areas were rare. Estrogen treatment resulted in a reduction of nuclear area and mitotic index, a changed nuclear shape, and an increased apoptotic index compared to that in vehicle-treated tumors. By castration, it is possible to induce an alteration of the androgen-sensitive Dunning R3327 PAP tumor phenotype to an androgen-insensitive tumor with an altered morphology. Estradiol-17 beta apparently inhibits not only the growth, but also postpones the castration-induced dedifferentiation of the tumor.

Computer-assisted chromatin texture characterization of Feulgen-stained nuclei in a series of 331 transitional bladder cell carcinomas

The chromatin patterns of Feulgen-stained nuclei in a series of six specimens of normal mucosa and 331 transitional bladder carcinomas, including 293 superficial (Ta and T1) and 38 invasive (T2-T4) cases, were quantitatively described by means of eight parameters relating to densitometric, run-length distribution, and co-occurrence matrix features. The results show that the chromatin texture of the superficial lesions was markedly different from that of the invasive tumours, which exhibited a distinctly more dense and heterogeneous chromatin pattern. The data also show that the increasing level of malignancy, as revealed by the increasing clinical stage, was accompanied by an increase in the overall chromatin condensation level. Only some areas of the nucleus actually increased in density; other pale areas appeared concomitantly with these increasingly denser chromatin areas. This chromatin density increase corresponded to a marked increase in the frequency of small dense chromatin clumps; these joined together into very large dense chromatin clumps, which were distributed more and more heterogeneously in the nucleus as the clinical stage of the tumour increased.

Pre- and postoperative DNA ploidy patterns correlated to pT-stage, histological grade and tumour volume in total prostatectomy specimens

Fifty-four consecutive patients with a preoperative diagnosis of localized prostatic cancer underwent total retropubic prostatectomy. The specimens were step-sectioned at 5 mm intervals. Volumes of invasive cancer (IC) and prostatic intraepithelial neoplasia (PIN) were calculated by computerized planimetry. From preoperative core and fine needle aspiration biopsies and from each prostatectomy specimen, multiple samples were taken from IC and PIN areas for DNA ploidy analysis. The study aimed to evaluate current biopsy sampling techniques as regards their suitability for DNA analysis and to assess the heterogeneity of DNA-ploidy and correlate the latter to postoperative pT-stage and tumour volume. When compared with the prostatectomy specimens, the preoperative assessment of non-diploid DNA patterns in biopsies had a sensitivity of 62% and a specificity of 86%. Samples from 24 surgical specimens contained non-diploid DNA in the main tumour and in one or several of the satellites. However, all these cases also contained diploid cell lines, both in the main tumour and in one or several of the satellites. Tumours with a volume exceeding 12 cc:s were non-diploid in 87.5% of cases (7/8). Tumours with volumes between 8 and 12 cc:s contained non-diploid foci in 50% of cases (2/4). Tumours smaller than 2 cc:s were non-diploid in 18% of cases (2/11). All non-diploid tumours were moderately or poorly differentiated. Of the non-diploid tumours, 96% (23/24) displayed capsular penetration versus 57% (17/30) of the diploid tumours (p < 0.0001). All PIN samples were made up of diploid cell lines.(ABSTRACT TRUNCATED AT 250 WORDS)

Karyometry in recurrent superficial transitional cell tumors of the bladder

Transitional cell carcinoma of the bladder has a high recurrence rate after local treatment. Progression to a higher stage occurs in 10-30% of the recurrent tumors, and early detection of potentially progressive tumors is important. In the current study morphometric, densitometric, and chromatin textural features of nuclei of superficial bladder tumors (pTa-T1) were studied to determine the value of karyometric features in the prediction of tumor progression. Seventy-two histological samples from 36 patients, consisting of both the primary and the first recurrent superficial tumor, were analyzed. Patients were divided into two groups: those with tumor progression, defined as an increase in tumor stage or occurrence of metastatic disease, and those without. Discriminant analysis on four karyometric features resulted in correct prediction of prognosis of 78% and 97% in the primary and recurrent tumors, respectively (P < 0.001). Tumor grade and stage did not offer additional information concerning prognosis. Karyometric analysis of recurrent superficial transitional cell tumors can be useful in selecting patients who need a more aggressive therapy. However, tumor characteristics of recurrent tumors varied and continuous evaluation of the karyometric features is necessary for early detection of an increase in the malignant potential of the tumor.