Human health studies to assess the effects of environmental exposures often make use of laboratory tests for so-called “biomarkers” of exposure, susceptibility and health effects (1). These tests are especially prominent components in studies of the immune response and host defense system (2).  Peripheral blood provides a convenient source of humoral immune mediators such as immunoglobulins (antibodies), and of certain immune cells including lymphocytes. Two distinct types of lymphocytes, T-cells and B-cells, are responsible for antigen-specific immune recognition. Stimulated B-cells are in turn responsible for producing antibodies.

    Lymphocyte phenotype (LPT) analysis by flow cytometry can distinguish T-cells and B-cells, as well as other lymphocyte subsets, on the basis of their surface receptor proteins. In the course of LPT analysis on several thousand samples from health assessment studies conducted by the Agency for Toxic Substances and Disease Registry (ATSDR), we noticed eleven samples with
characteristics seen in the malignant B-cells of chronic lymphocytic leukemia (B-CLL). This report describes the laboratory findings in those eleven individuals.
 

    Lymphocyte phenotypes (LPT) were determined by methods generally consistent with established guidelines for clinical analysis and in compliance with regulations of the Clinical Laboratory Improvement Act (CLIA).

    Whole blood samples were prepared for analysis by a stain-and-lyse method (see Table 2) using two monoclonal antibody conjugates in each tube (Table 1), one labeled with fluorescein isothiocyanate (FITC) and the other with phycoerythrin (PE). All samples were analyzed with a minimum of the 6-tube “basic” panel. In a subset of participants, additional tubes were
included in an “extended” panel to measure lymphocytes bearing CD5 or HLA-DR. Because these tests were in a technical research and development phase, results from them were not reported to study participants.

    Stained sampled were analyzed on either a EPICS 741 or EPICS Elite flow cytometer (Coulter Corporation, Hialeah, FL) using the 4Cyte independent 8 bit  4-parameter acquisition system and Acmecyte customized software (3).  Both cytometers were calibrated using consistent target conditions for light scatter, fluorescence intensity, and spectral compensation parameters (4).  Data were acquired and archived in “listmode” files containing all information
recorded during analysis (5).

    The LPT percentages were determined by quadstat analysis of fluorescence distributions of events defined as lymphocytes by forward and right angle light scatter. A rectangular light scatter gate determined by the analyst during data acquisition was used to delineate lymphocytes. Quadstat cursor positions, used to dichotomize events into “negative” and “positive” populations
on both the x and y axis, were determined by inspection for most phenotypes, which had clear separation between negative and positive distributions (see example Figure 2). Since the B-cells lacked a clear separation between CD5-positive and CD5-negative events, a non-specific fluorescence (NSF) control (see Table 1, footnote 3 for operational definition of NSF) for each
sample was run separately and used to determine the quadstat cutoff point for CD5 (see Figure 6). The same approach was used to discriminate HLA-DR staining on T-cells.
 

    LPT were reported from EHLS/CDC only as percentages of lymphocytes.  Complete blood counts (CBC), performed in different laboratories depending on the location of the study, were used to determine the total lymphocyte counts, which were multiplied by the LPT percentages to obtain total (absolute) LPT counts. Although different methods were used to perform the CBC
at different sites, review of leukocyte and lymphocyte counts did not reveal any notable biases between sites referable to methodology.
 

    Of the approximately 6000 samples analyzed for ATSDR studies, about 30 repeat drawings were requested. In addition to the eleven atypical B-cell LPT described below, these included some samples with very low (<20%) CD4 percentages, some with very high (>30%) natural killer (NK) cell percentages, and a few in which sample integrity had apparently been compromised.
 

    Most measurements for IgG, IgA, and IgM (including 10 of the 11 cases) were performed at the Foundation for Blood Research by laser turbidometry.  All results were calibrated to the U.S. National Reference Preparation maintained at the Centers for Disease Control and Prevention (6).
 

    EHLS/CDC analyzed about 6000 blood samples from ten ATSDR studies from 1991 through 1994. In the course of these studies, analysts identified samples from 11 participants with phenotypic characteristics that were similar to those seen in early B-cell chronic lymphocytic leukemia (B-CLL). These identifications were made subjectively, based upon the analysts' knowledge of normal distributions and staining patterns for white blood cell surface markers by flow cytometry. At the time of identification, the analysts were unaware of the total cell counts or demographic information other than age. Because of their phenotypic similarity to early B-CLL, the samples were designated “B-CLL-like LPT”.

    These phenotypic characteristics are depicted in Figures 1-6. The major features observed in varied combinations among the eleven cases were: altered light scatter (representing morphologic alterations); a high percentage of B-cells; dim staining for the CD45 marker; dim staining for the CD20 marker; and bright staining for the CD5 marker on B-cells identified by CD19. A case-by-case summary of findings is given in Table 3.
 

    LPT are determined using fluorescent stains conjugated to antibodies that bind specifically to cell surface receptors. A minimum of about 1000 molecules of bound antibody is required to distinguish a labeled lymphocyte from background, making it “positive” for the particular marker recognized by that antibody. Typical analysis is limited to determining the percentage of lymphocytes that is positive for a particular phenotype. However, with proper calibration,
the number of bound antibody molecules per cell can be determined from the fluorescence intensity (7).
 

    Because of the great variety of LPT distributions, the array of quadstat displays from a comprehensive LPT analysis on a blood sample is almost like a fingerprint. Experienced analysts can quickly recognize unusual quadstat patterns.  Analysts at EHLS/CDC were particularly familiar with the wide ranges of normal variation, because they had analyzed samples from more than 6000 persons free of diseases that effect LPT. This experience is exceptional, since LPT is far too complex to be used as a routine screening tool. Almost all LPT on human samples in both research and clinical settings is devoted to either HIV-related conditions or to leukemias and lymphomas. While some reference range studies have been conducted in normal populations, these studies have examined at most a few hundred samples, in contrast to the several thousand
analyzed at EHLS/CDC.

    An atypical LPT pattern suggestive of B-CLL is exemplified by CD45 staining in Cases 1, 5, and 6, which showed a distinct double population of CD45-stained lymphocytes, with a less-brightly stained population to the left of the normal staining position (Figure 2). CD45 is expressed by all leukocytes, but normal lymphocytes stain more brightly for CD45 than on any other leukocytes. Reduced CD45 staining is known to be characteristic of some types of B-CLL. These three B-CLL-like LPT cases are the only examples of noticeable diversity in CD45 staining among the 6000 samples analyzed at EHLS/CDC.

    The CD20 staining (Figure 4) also provides several examples of clearly abnormal LPT patterns suggestive of B-CLL. CD20 is expressed strongly on mature peripheral blood B-cells, which normally account for less than 25% of peripheral blood lymphocytes in adults. In B-CLL patients, B-cells can account for almost all peripheral blood lymphocytes, and the B-CLL cells themselves often stain only weakly for CD20. Case 1 shows not only a very high (67%) proportion of B-cells, and it also shows the majority of them with weak-to-moderate staining (a small number of normally-bright CD20 cells can be seen above the main cluster). A similar pattern is seen in Case 6. Dichotomous CD20 staining is especially pronounced in Case 11, where roughly equal
populations of normally- and weakly-staining CD20 B-cells are evident. The same type of pattern is seen in Case 10, even though the proportion of B-cells (12%) is well within the normal range. In Case 5, CD20 staining is so dim that the B-cell cluster merges with the unstained cells, so that CD20 no longer provides a valid marker for counting B-cells.

    Perhaps the most compelling changes in LPT related to B-CLL were seen with the CD5 receptor. CD5 is normally expressed at high levels only on T-cells, while it is absent or only weakly expressed on normal B-cells. In fact, expression is so weak that LPT assays may not distinguish CD5 staining from non-specific background on normal B-cells. The assay used at EHLS/CDC was designed with a separate non-specific control to insure proper discrimination.
The resulting pattern (Figure 5) normally shows bright staining for T-cells and a streak-like cluster of B-cells extending from the CD5-negative region into the weakly positive region. A clear departure from this pattern was observed in all of the 7 cases where CD5 LPT were analyzed. Cases 4, 7, 10 and 11 all show a bright CD5 cluster among the B-cells, along with a small number of presumably-normal B-cells with weak or absent CD5 staining. Case 8, from a participant previously-diagnosed with B-CLL, shows all of the B-cells staining brightly for CD5. Case 5 shows an amorphous cluster with most of the B-cells staining positive for CD5. Case 9 shows an amorphous cluster with most of the B-cells negative for CD5.
 

    The 11 individuals described in this report as having “B-CLL-like LPT” were culled from all other samples because analysts noticed distinct variations in the LPT patterns that were suggestive of early B-CLL. Table 3 summarizes the features of these 11 cases. No formal case definition was established a priori, so the B-CLL-like LPT presented here must be considered as individual reports based on analysts' subjective assessment. Since analysts were not aware of the total leukocyte counts, the only enumerative information available to them was the percentage of B-cells among all lymphocytes.

    Retrospective data review shows that 8 of the 11 cases could be identified objectively as samples with B-cell percentages above the 99.4th percentile of the distribution among persons age 45 or greater, and this marked elevation was certainly noted at the time of analysis. However, even if the B-cell proportion had not been enumerated, changes in the LPT staining patterns described above would have led to classification as B-CLL-like LPT in all but one case:  only Case 3 had a preponderance of B-cells (48% of lymphocytes, 99.5th percentile of the distribution) with apparently-normal staining patterns, and CD5 was not assessed on this sample. Conversely, only one case failed to show an elevated proportion of B-cells: Case 10 was in the mid-range of the B-cell percentage distribution, but it showed dichotomous CD20 staining as well as a distinct CD5-bright cluster among the B-cells. Thus, 9 of the 11 cases showed both an elevated proportion of B-cells and LPT staining patterns suggestive of B-CLL.
 

    The recognition of B-CLL-like LPT was based on certain changes in LPT staining patterns recognized during sample analysis. One of the 11 cases came from a person diagnosed with CLL, and three others showed monoclonal B-cell populations by kappa-lambda analysis (Chapter 3). These findings suggests that the presence of B-CLL-like LPT reflects increased risk for B-cell lymphoproliferative disorders.