Brian H. Mayall

Guidelines for the implementation of clinical DNA cytometry

SummaryThese consensual guidelines and recommendations address the potential utility of DNA cytometry in characterizing human malignancies. They are provided to inform laboratory personnel, pathologists, and clinicians about DNA cytometry. For individual patients, use of DNA cytometry, selection of specific techniques, and interpretation and utilization of results remain the responsibility of the attending physicians.

Quantification and classification of human sperm morphology by computer-assisted image analysis.

A quantitative, semi-automated method for classifying human sperm based on objective measurements of head shapes and sizes has been developed. Air-dried smears of semen from eight healthy men were stained with the Feulgen reaction and 283 sperm were selected as prototypic examples of the 10 morphology classes used in our classification system. Sperm heads were imaged through a microscope (NA = 1.3), sampled at 0.125-micron intervals, and measured on an image analysis system. Measurements included stain content, length, width, perimeter, area, and arithmetically derived combinations. Additionally, each sperm image was optically sectioned at right angles to its major axis to give a measure of lengthwise heterogeneity of shape. Linear stepwise discriminant analysis was used to identify the more powerful parameters and to create a model employing eight parameters. The jackknifed classification procedure distinguished normal from abnormal sperm with 95% accuracy and correctly assigned 86% of the sperm to one of 10 shape classes. Most of the misclassification errors occurred among closely related classes. The results demonstrate the ability of automated image analysis to classify individual sperm into clinically familiar shape categories.

Comparison of five histopathologic methods to assess cellular proliferation in transitional cell carcinoma of the urinary bladder

Transitional cell carcinomas of the urinary bladder vary in their biologic potential, which may be correlated with the grade and stage of the tumor. Cellular proliferation may prove to be another measure of predicting tumor biologic potential. We have compared five different methods to assess proliferation in 26 tumors and correlated these results with tumor grade and stage. A portion of each tumor was incubated in vitro with bromodeoxyuridine (BrdUrd). For each tumor this was compared with at least three of the following four other markers of proliferation: mitotic count, silver-stained nucleolar organizer regions, immunohistochemical staining with Ki67, and proliferating cell nuclear antigen. Statistical correlations were seen between tumor grade and stage and these markers. There were strong correlations between the BrdUrd labeling index (LI) and both the Ki67 LI and proliferating cell nuclear antigen LI. The correlation between the BrdUrd LI and mitotic count was more tenuous; no significant correlation was found between BrdUrd LI and silver-stained nucleolar organizer region count. The correlation between these measurements of proliferation and tumor grade and stage was less strong. Our data suggest that cellular proliferation of transitional cell carcinomas can be reliably assessed with several different markers and that most of these markers can be correlated with tumor grade but not with stage.

DEOXYRIBONUCLEIC ACID CYTOPHOTOMETRY OF STAINED HUMAN LEUKOCYTES II. THE MECHANICAL SCANNER OF CYDAC, THE THEORY OF SCANNING PHOTOMETRY AND THE MAGNITUDE OF RESIDUAL ERRORS

The mechanical scanner of CYDAC is a dual beam cytophotometer with a Nipkow scanning disc, analogue photometric circuits, digital conversion and accumulation of the photometric values and digital control of the scanning procedure. A measuring sequence involves object and clear field scans, and the final readout is virtually independent of fluctuations and heterogeneities in illumination and of variations in the size of the scanning apertures. Optical sources of error are analyzed both theoretically and experimentally as they apply to scanning cytophotometers. The effects of conical and polychromatic illumination, stray light and heterogeneous distribution of chromophore lead to negative averaging errors that are a function of optical density. The role of scan aperture size and optical resolution on distributional error is discussed and calculated in terms of three possible models. Non-absorptive phenomena, such as scattering and diffraction, lead to positive errors that are complementary, in a qualitative sense, to distributional error. In attempting to isolate these different errors experimentally, we show that measurements of the relative deoxyribonucleic acid stain content of human leukocytes are independent, within wide limits, of condenser and objective numerical aperture and of changes in absorptivity secondary to changes in wavelength. Where an adequate theory exists, these experimental results agree well with the theoretical predictions and, in any case, they indicate that for stained leukocytes the net effect of all optical errors must be less than 2%. Random measuring errors include both photon shot noise and errors associated with the geometry of the scanning disc. These errors contribute to an error of 1.2% among replicate measurements of leukocytes made under standard conditions. The previously reported differences in stain content among leukocyte types and among individual cells within types cannot be ascribed to either optical or instrumentation errors. Thus, the measured differences must reflect actual differences either in the amount or in the absorptivity of the stain.

Purification of the chromosomes of the Indian muntjac by flow sorting.

Metaphase chromosomes were isolated from a male Indian muntjac cell line, were stained with ethidium bromide and were analyzed by flow microfluorometry to establish a deoxyribonucleic acid (DNA)-based karyotype. Five major peaks were evident on the chromosomal DNA distribution corresponding to the five chromosome types in this species. The amount of DNA in each chromosome was confirmed by cytophotometric measurements of intact metaphase spreads. The five chromosome types were separated by flow sorting at rates up to several hundred chromosomes per second. The sorted chromosomes were identified by morphology and by Giemsa banding patterns. The automsomes, Numbers 1, 2 and 3, and the X + 3 composite chromosome were separated with a high degree of purity (90%). The centromere region of the X + 3 chromosome was fragile to mechanical shearing, and during isolation a small proportion of these chromosomes broke into four segiments: the long arm, the short arm, the short arm plus centromere and the centromere region. A large fraction of the constitutive heterochromatin of this species is present in the centromere region of the X + 3 chromosome and in the Y chromosome; these two regions possess similar amounts of DNA and therefore sort together. Chromosome flow sorting is rapid, reproducible and precise; it allows the collection of microgram quantities of purified chromosomes.

The prognostic value of proliferation indices: a study with in vivo bromodeoxyuridine and Ki-67

Proliferation indices are intended to help patients and clinicians make treatment decisions. We have previously demonstrated that a proliferation index based on in vivo labeling of S-phase cells with bromodeoxyuridine (BrdUrd) correlates with Ki-67 labeling index (LI). We now compare the prognostic value of these indices.With written consent, we gave 129 women with biopsy confirmed breast cancer 200 mg/M2 BrdUrd during 30 min immediately preceding surgery. We used IU-4 anti BrdUrd antibody to count the immunohistochemical labeling index (LI) of DNA-incorporated BrdUrd in 2,000 cells and MIB-1 to count Ki-67 (118 cases). Patients received standard surgical and adjuvant treatment. No patients were lost to follow-up and patients were followed a minimum of 2 (median 5.1) years. We compared survival and recurrence in tumors with high vs low labeling indices. We found that women in the low BrdUrd LI group had better disease free survival (92% vs 67% 5-yr DFS p = 0.001) and overall survival (94% vs 70% 5-yr OS, p = 0.0001) than those with a high LI. In comparison, a low Ki-67 index predicted better OS (87% vs 80% 5-yr OS, p = 0.020) and a trend for better DFS (84% vs 72% DFS p = 0.055). The apparent superiority of BrdUrd LI over Ki-67 LI is likely due to chance (p = 0.18). In multivariate survival analyses we found that BrdUrd LI proliferative index significantly improves prediction of DFS or OS even when node status, age or tumor size is in the model. We conclude that markers of proliferation are useful adjuncts in predicting patient prognosis.

Image cytometric classification of premalignant breast disease in fine needle aspirates

Image cytometry for the classification of fine needle aspirate (FNA) biopsies was evaluated in samples from 39 women. Eighteen of them had benign lesions, seven had premalignant lesions, nine had carcinoma in situ and five had carcinoma. The term, premalignant, here refers to lesions with an increased risk of developing into breast cancer (atypical hyperplasia and, to a lesser extent, moderate or florid hyperplasia). The classifications by cytometry were compared with the microscopic diagnoses of the same FNA samples and of tissue from a subsequent surgical biopsy of the same area. One slide from each breast FNA sample was restained in Azure‐A Feulgen. Breast epithelial cells were measured using a texture analysis program on the Leitz TAS‐plus. The mean, standard deviation (SD), and interquartile range were calculated for each of 12 nuclear parameters from 200 cells per slide. A discriminant analysis was used to develop a statistical model for classifying individual samples. Six of seven atypical proliferative lesions (atypical hyperplasia and moderate hyperplasia) were identified by image cytometry, but were unrecognized by conventional microscopic examination.

Cell Proliferation in Prostatic Adenocarcinoma: In Vitro Measurement by 5-Bromodeoxyuridine Incorporation and Proliferating Cell Nuclear Antigen Expression

We assessed cancer cell proliferation, a marker of the biologic activity of tumor cells, by evaluating bromodeoxyuridine (BrdUrd) incorporation and proliferating cell nuclear antigen (PCNA) expression. Prostatic carcinoma specimens (N = 48) were incubated in the presence of BrdUrd to label cells undergoing DNA synthesis, and immunocytochemical staining was performed with monoclonal antibodies to BrdUrd and PCNA and a standard indirect immunoperoxidase technique. The proportion of cells staining positively (labeling index or LI) for BrdUrd and PCNA was determined in 2 ways: by counting only high-power fields with the greatest concentration of stained cells (selected LI); or by counting cells in random fields (random LI). For BrdUrd the mean selected and random LIs were 3.08% and 1.62%, respectively; for PCNA they were 6.02% and 3.47%. Random and selected BrdUrd correlated well (r2 = 0.83), as did random and selected PCNA LIs (r2 = 0.86). However, a weaker correlation was noted when LIs of both techniques were compared, with the PCNA LI usually higher. The LIs of either technique correlated rather poorly with tumor grade and concentration of prostate-specific antigen, but correlated well with clinical stage as assessed by examination and imaging. In addition, either technique discriminated among tumors known to be pathologically confined (stages A and B) and those with extension to seminal vesicles (stage C) or metastatic to regional lymph nodes or bone (p < 0.019).

STRATEGIES FOR CHOOSING A DEOXYRIBONUCLEIC ACID STAIN FOR FLOW CYTOMETRY OF METAPHASE CHROMOSOMES

Requirements for flow cytometry of metaphase chromosomes stained with three deoxyribonucleic acid (DNA)-specific fluorescent dyes--Hoechst 33258, Chromomycin A3, and ethidium bromide--are reviewed. Fluorescence properties of these three stains when bound to mitotic cells or to chromosomes in suspension are measured and compared with fluorescence properties when bound to DNA in solution. Conditions are given for high resolution flow cytometry of Chinese hamster chromosomes stained with each of the fluorophors, and histograms are presented that exhibit differences in relative peak position and area. Energy transfer fluorescence between two DNA stains is presented as a potentially useful new parameter for flow cytometry of chromosomes and is illustrated by fluorescence energy transfer from Chromomycin A3 to ethidium bromide when simultaneously bound to hamster mitotic cells.

Central giant cell granulomas of the jaws: Nuclear DNA analysis using image cytometry

Giant cell nuclear DNA, in 30 giant cell lesions of the jaws, was quantified by computer-assisted image analysis. DNA content was then used to predict clinical behavior and outcome. 4 nuclei in each of 25 giant cells (total = 100 nuclei) were randomly selected and the DNA content was quantified by the Leitz Texture-Analysis-System-Plus. DNA in nuclei of normal appearing stromal fibroblasts (n = 20) was similarly measured. The DNA index was calculated as the mean nuclear DNA content of giant cells divided by the mean DNA content of control fibroblasts. The mean DNA-index of aggressive lesions (1.09, SD = 0.12) was not significantly different from that of non-aggressive lesions (1.18, SD = 0.15) (p = 0.093). The results indicate that the nuclear DNA content of giant cells is not useful as a predictor of the clinical behavior of giant cell lesions of the jaws.