Effects of resolution reduction on data analysis.

BACKGROUND: There is often a need in flow cytometry to display and analyze histograms at resolutions lower than those native to the data. It is common, for example, to analyze DNA histograms at 256-channel resolution, even though the data were acquired at 1,024 channels or more. The most common method for reducing resolution, referred to as the consecutive summation (CS) method, can introduce distortions into the shape of histograms. Peaks that were symmetric in the original data can become skewed in the reduced-resolution histogram. Data analysis can be negatively affected by the distortions produced by reducing the histogram resolution. An alternative technique for reducing histogram resolution, the unbiased summation (US) method, minimizes shape distortion. This paper describes the US method and examines the benefits it provides in the analysis of DNA histograms. METHODS: Reduced chi-square (RCS) was used to measure the response to three experimental variables in the least-squares analysis of simulated DNA histograms. For each variable (the percentage of coefficient of variation [%CV], number of events, and mean position of the G1 distribution), a test data set of 1,000 histograms was generated at 1,024-channel resolution. Histogram resolutions were reduced with each method and then analyzed with ModFit LT cell-cycle analysis software (Verity Software House, Topsham, ME). S-phase error and processor computation time of each method also were evaluated. A Monte Carlo experiment was performed to compare CS and US methods to theoretically correct reductions. RESULTS: CS method analysis results were negatively affected by changes in %CV, number of events, and G1 peak position. The US method produced consistently lower RCS values (more accurate results) within the tested ranges. The US method eliminated bias in S-phase error and had negligible impact on analysis processing speed. It improved RCS values 44.50% on average (P < 0.0002) with actual DNA histograms. Whereas the CS method became less accurate (chi-square test) as the amount of reduction increased, the US method was unaffected, producing consistently better results. CONCLUSIONS: The US method is recommended for reducing histogram resolution in modeling applications such as DNA cell-cycle analysis. It may have implications in other areas of flow cytometric data analysis.

Optimizing flow cytometric DNA ploidy and S-phase fraction as independent prognostic markers for node-negative breast cancer specimens.

Developing a reliable and quantitative assessment of the potential virulence of a malignancy has been a long-standing goal in clinical cytometry. DNA histogram analysis provides valuable information on the cycling activity of a tumor population through S-phase estimates; it also identifies nondiploid populations, a possible indicator of genetic instability and subsequent predisposition to metastasis. Because of conflicting studies in the literature, the clinical relevance of both of these potential prognostic markers has been questioned for the management of breast cancer patients. The purposes of this study are to present a set of 10 adjustments derived from a single large study that optimizes the prognostic strength of both DNA ploidy and S-phase and to test the validity of this approach on two other large multicenter studies. Ten adjustments to both DNA ploidy and S-phase were developed from a single node-negative breast cancer database from Baylor College (n = 961 cases). Seven of the adjustments were used to reclassify histograms into low-risk and high-risk ploidy patterns based on aneuploid fraction and DNA index optimum thresholds resulting in prognostic P values changing from little (P < 0.02) or no significance to P < 0.000005. Other databases from Sweden (n = 210 cases) and France (n = 220 cases) demonstrated similar improvement of DNA ploidy prognostic significance, P < 0.02 to P < 0.0009 and P < 0.12 to P < 0.002, respectively. Three other adjustments were applied to diploid and aneuploid S-phases. These adjustments eliminated a spurious correlation between DNA ploidy and S-phase and enabled them to combine independently into a powerful prognostic model capable of stratifying patients into low, intermediate, and high-risk groups (P < 0.000005). When the Baylor prognostic model was applied to the Sweden and French databases, similar significant patient stratifications were observed (P < 0.0003 and P < 0.00001, respectively). The successful transference of the Baylor prognostic model to other studies suggests that the proposed adjustments may play an important role in standardizing this test and provide valuable prognostic information to those involved in the management of breast cancer patients.

DNA and cell cycle analysis as prognostic indicators in breast tumors revisited.

Both DNA ploidy and S-phase ploidy are promising prognostic factors for node-negative breast cancer patients. Based largely on the analysis of one large study, much of the reported problems with these factors have been caused by some unappreciated complexities in categorizing DNA ploidy into low- and high-risk groups and the lack of some necessary adjustments to eliminate unwanted correlations between DNA S-phase and ploidy. When both DNA ploidy and S-phase are compensated properly, they become independent prognostic markers, forming a powerful prognostic model.

Data analysis through modeling.

UNLABELLED: Flow cytometers collect data in an attempt to understand some aspect of a biological system. Mathematical models are developed to extract relevant population features from the data, to help users attain their goals. This unit examines what models are and how they are routinely used. KEYWORDS: flow cytometry; listmode data; data analysis; modeling Flow cytometers collect data in an attempt to understand some aspect of a biological system.

Proposed new data file standard for flow cytometry, version FCS 3.0.

In 1984, the first flow cytometry data file format was proposed as Flow Cytometry Standard 1.0 (FCS1.0). FCS 1.0 provided a uniform file format allowing data acquired on one computer to be correctly read and interpreted on other computers running a variety of operating systems. That standard was modified in 1990 and adopted by the Society of Analytical Cytology as FCS 2.0. Here, we report on an update of the FCS 2.0 standard which we propose to designate FCS 3.0. We have retained the basic four segment structure of earlier versions (HEADER, TEXT, DATA and ANALYSIS) in order to maintain analysis software compatibility, where possible. The changes described in this proposal include a method to collect files larger than 100 megabytes (not possible in earlier versions of the standard), the inclusion of international characters in the TEXT portions of the file, a method of verifying data integrity using a 16-bit cyclic redundancy check, and increased keyword support for cluster analysis and time acquisition. This report summarizes the work of the ISAC Data File Standards Committee. The complete and detailed FCS 3.0 standard is available through the ISAC office [Sherwood Group, 60 Revere Drive, Ste 500, Northbrook, IL 60062, phone: (847) 480-9080 ext. 231, fax: (847) 480-9282, E-mail: isac@sherwood-group.com] or through the internet at the ISAC WWW site, http://nucleus.immunol.washington.edu/ISAC.ht ml.

Parathyroid carcinoma: the relationship of nuclear DNA content to clinical outcome.

This article reports the use of flow cytometry to determine tumor nuclear DNA content and its correlations with clinical outcome in a series of patients with parathyroid carcinoma. Information concerning nine patients with parathyroid cancer (aged 25 to 88 years) was reviewed. Paraffin-embedded, formalin-fixed archival tissue was used to determine tumor DNA content flow cytometrically. Twenty-five operative procedures were performed in nine patients, including 11 parathyroidectomies, two wide local excisions, six central neck dissections, and four median sternotomies for resection of metastases. With flow cytometry used to determine a tumor DNA index, five patients had evidence of tumor aneuploidy; in two patients two aneuploid peaks were evident. The DNA index ranged from 0.7 (hypodiploid) to 1.92 (mean, 1.31). Follow-up ranged from 1 to 18 years. Four patients died. Five were alive 1 to 13 years after diagnosis of parathyroid disease. Four of the five patients with evidence of tumor aneuploidy had metastatic disease and died, and the fifth has had three local recurrences. The four patients with diploid tumors were alive and free of disease 1, 3, 4, and 8 years after the initial operation. It is concluded that in patients with clinically or pathologically demonstrated parathyroid cancer, flow cytometry may help differentiate those whose cancers are likely to behave indolently (diploid tumors) from those with tumors (aneuploid) more likely to behave aggressively by recurring locally or metastasizing.

Nuclear DNA analysis of hyperplastic parathyroid glands in multiple endocrine neoplasia type I.

Twenty-four hyperplastic parathyroid glands from 11 patients with multiple endocrine neoplasia type I (MEN-I), and 36 hyperplastic parathyroid glands in 15 patients with sporadic primary hyperparathyroidism, ie, not associated with MEN, were analyzed for DNA by flow cytometry. Sixteen of 24 hyperplastic parathyroid glands from patients with MEN-I were DNA diploid, and eight were DNA aneuploid. Thirty-three of 36 hyperplastic parathyroid glands from patients without MEN were DNA diploid, and only three were DNA aneuploid. The mean percentage of 4c level (a measure of the G2M phase of the cell cycle) of DNA diploid hyperplastic parathyroid glands taken from patients with MEN-I was 8.1% +/- 4.5%, which is significantly higher than the 3.5% +/- 3.4% for those taken from patients without MEN. Our results show that there is a difference in nuclear DNA content between hyperplastic parathyroid glands in patients with MEN-I and those in patients without MEN.

Single and multigland disease in primary hyperparathyroidism: clinical follow-up, histopathology, and flow cytometric DNA analysis.

Two-hundred seventy-four patients with primary hyperparathyroidism had selective removal of enlarged parathyroid glands. Biopsies were taken from all parathyroid glands. Normal-size glands were not resected irrespective of their histological appearance. After a mean follow-up of 13.5 years the rates of persistent and recurrent hyperparathyroidism were, respectively, 3.6% and 0.7%. Transient and permanent hypoparathyroidism occurred in 24% and 2.5% of the patients. The microscopic appearance of enlarged glands and of biopsies taken from normal-size glands were reviewed by two pathologists. Normal parathyroid glands were distinguished from abnormal glands fairly accurately (sensitivity 93%, specificity 80%). Microscopic classification of abnormal parathyroid glands as adenomas or hyperplastic glands correlated poorly with the gross classification as single or multigland disease. Flow cytometric DNA analysis of paraffin embedded parathyroid tissue showed significant differences for DNA index, % S-phase and % G2M (p less than 0.001). Differentiating single from multigland disease by means of DNA analysis was not possible. In conclusion, removal of only enlarged parathyroid glands results in acceptable rates of persistent and recurrent hyperparathyroidism. Biopsies should only be taken sparingly to prevent transient and permanent hypoparathyroidism. Microscopic examination and flow cytometric DNA analysis can differentiate normal from abnormal parathyroid glands but are unable to differentiate abnormal glands into single or multigland disease.

DNA signal splitting improves detection and analysis of tetraploid populations.

Detection of DNA tetraploid populations requires a high index of suspicion at the time of data acquisition and frequently requires subsequent off-line analysis for confirmation, including evaluation of the hypertetraploid region. To analyze these specimens, the flow cytometer operator must run all specimens with the G0G1 peak in low channels or rerun specimens in which tetraploidy is suspected with a lower photomultiplier tube (PMT) voltage or lower amplifier gain setting. Re-analysis may not be possible in specimens with few cells. A simple modification to the cytometer allows PMT signal splitting with simultaneous processing of the signal by two different amplifiers. This allows simultaneous acquisition of histograms optimized for both the hypotetraploid and hypertetraploid regions.

DNA histogram debris theory and compensation.

In this paper we describe a theory of DNA histogram debris generation and compensation that can be applied to paraffin-embedded frozen tissue preparations. The theory predicts the distribution of fragments generated from single and multiple random sectioning of three-dimensional ellipsoids representing nuclei. The fragment distribution is assumed to be a major component of the underlying debris in DNA histograms. A comparison of S-phase fractions (SPF) from matched tissue prepared by frozen and formalin-fixed paraffin-embedded DNA methods demonstrates the usefulness of the theory.

Improved flow cytometric determination of proliferative activity (S-phase fraction) from paraffin-embedded tissue.

Recent studies suggest that proliferative activity (S-phase fraction [SPF]) may have greater prognostic significance than total nuclear DNA content; however, relatively few studies have examined SPF from paraffin-embedded tissue because of significant contamination of histograms with debris. In this study, cell cycle analysis was performed on 124 matched tissue specimens. Fresh tissue was divided into two equal portions; one portion was frozen, whereas the other portion was processed and embedded in paraffin. S-phase could be determined for both frozen and paraffin-embedded tissue in 81 cases. Correlation between SPF from frozen and paraffin-embedded tissue was demonstrated (r = 0.80) when debris was subtracted from histograms with the use of two new subtraction algorithms referred to as multicut and singlecut. Unlike other debris-subtraction algorithms, the quantity and distribution of debris calculated by these algorithms are dependent on the magnitude and position of histogram peaks. A lesser degree of correlation was demonstrated with the use of a standard exponential debris subtraction algorithm (r = 0.67). Correlation of SPF for aneuploid cases was greater when SPF was calculated as a percentage of the aneuploid cell population rather than as a percentage of the entire cell population. This was attributed to the observation that the proportion of aneuploid cells from paraffin-embedded tissue was less than that from frozen tissue. The results of this study indicate that SPF can be calculated from paraffin-embedded tissue with values comparable to those obtained from frozen tissue. The ability to calculate SPF reliably from paraffin-embedded tissue should allow additional evaluation of this parameter as a prognostic indicator.

Introduction to flow cytometry data file standard.

The Data File Standards Committee of the Society for Analytical Cytology presents a Standard to be used for the storage of data associated with flow cytometric measurements. The Standard specifies a format that provides for the inclusion of all information necessary to fully describe: 1) the instrument used for the measurement; 2) the sample measured; 3) the data obtained; and 4) the results of analysis of the data. The Committee and the Society for Analytical Cytology point out that the use of this Standard by all those individuals and companies that generate or use data taken with flow cytometers or generate methods of analysis for the data will encourage the sharing of such data and methods of analysis.

A simple and rapid method for determining the linearity of a flow cytometer amplification system.

We describe a simple and rapid method for determining the linearity of a flow cytometer amplification system. The method is based on a fundamental characteristic of linear amplifiers: The difference between two amplified signals increases linearly with increasing amplifier gain. Two populations of beads or cells, differing slightly in fluorescence intensity, are analyzed by the flow cytometer at increasing photomultiplier tube high-voltage settings. The distribution of the populations' mean difference versus mean position is a straight line intersecting the origin for linear amplifiers. Although some types of nonlinearities cannot be detected with this technique, deviations from linearity indicate nonlinear components in the flow cytometer amplification system. The correlation coefficient is used to quantify degree of nonlinearity. We also describe a method for amplifier nonlinearity compensation.

Effects of several commonly used fixatives on DNA and total nuclear protein analysis by flow cytometry.

Five commonly used fixatives (AZF, B-5, Bouin's, formalin, and Zenker's) were evaluated for their effect on the flow cytometric analysis of DNA and total nuclear protein (TNP) in solid tumors. Data were obtained with the use of colonic adenocarcinoma, squamous carcinoma of the lung, mammary adenocarcinoma, and spleen with a plasma cell leukemic infiltrate. The parameters examined were G0-G1 DNA staining intensity, %G0-G1, percent coefficient of variation (%CV), percent debris, and TNP staining intensity. The results showed that variations in the fixation of solid tumor significantly affected flow cytometric-derived parameters. In this study, paraffin-embedded tissue (PET) fixed in 10% (v/v) neutral buffered formalin (NBF) produced the best results, with a %CV below 4.7, whereas fixatives such as Zenker's and B-5 produced poor %CVs (above 6.0) or uninterpretable TNP and light scatter data. These data suggest that a portion of all tissue samples be fixed in NBF to allow for subsequent analysis by fixative-sensitive assays such as DNA in situ hybridization and flow cytometry.

Localization of monoclonal B-cell populations through the use of Komogorov-Smirnov D-value and reduced chi-square contours.

In this study three different flow cytometric analysis techniques are woven together into a single system that permits improved detection of small percentages of monoclonal B cells in a milieu of either normal blood leukocytes or bone marrow cells. This analysis is an extension of the concept of clonal excess, which is used to detect the presence of a tumor that is a clonal expansion of B cells expressing either kappa or lambda light chains. The technique also utilizes "multiple listmode processing," which is defined in this context as the simultaneous analysis of two or more listmode files that share one or more common parameters. This type of data structure enables the segmentation of two parameter light scatter displays into regions from which numerous kappa and lambda histograms subsequently can be analyzed for their respective Komogorov-Smirnov D-values or R-values (reduced chi-square value). The final technique makes use of a calculated parameter display system. Superimposed on the light scatter dot density plot are D-value or R-value contours. The contours target the location of the population that is abnormal, thus providing information for setting optimal bitmap gates for clonal excess studies, other phenotypic analyses, or cell sorting. In experiments using model systems, the sensitivity of this assay is estimated to be between 0.25% and 2.5%. The technique's distribution information and sensitivity may prove useful for staging, treatment monitoring, and relapse detection of B-cell leukemia and lymphoma. This application illustrates the potential of combining multiple listmode processing and calculated parameter display to expand the effective dimensionality of listmode data.