Flow Cytometric Monitoring of Human Immunodeficiency Virus-Infected Patients

Transcription

1 Flow Cytometric Monitoring of Human Immunodeficiency Virus-Infected Patients Simultaneous Enumeration of Five Lymphocyte Subsets CHENG-MING LIU, PH.D., KATHARINE A. MUIRHEAD, PH.D., SELMA P. GEORGE, MT(ASCP), AND ALAN L. LANDAY, PH.D. The utility of CD4 lymphocytes in monitoring disease progression and prognosis of human immunodeficiency virus (HI V)-infected patients is well established. We have modified a previously described antibody cocktail to provide complete lymphocyte subset analysis on 1-2-ML samples of whole blood. This method optimizes accuracy of CD4 lymphocyte assessments and provides simultaneous assessment of four other lymphocyte subtypes of interest in specimens with absolute lymphocyte counts as low as 3 X 1 6 /L. Lymphocytes are classified as T helpcr (CD3 + CD4 + ); T s ppressor (CD3 + CD8 + ); T null (CD3+CD4CD8", putative 7«T-cell receptor); B (CD19 + CD2 + ); or natural killer (CD3~CD16 + CD6 + ). The method positively discriminates against contamination of lymphocyte scatter gates by monocytes and unlysed erythrocytes and is compatible with a variety of cell preparation procedures. Increased accuracy of CD4 lymphocyte determinations and simultaneous identification of other lymphocyte subsets whose relationship to disease progression is under study make this an efficient and informative method for disease monitoring and evaluation of therapy in HIV-infected patients. (Key words: Lymphocyte subsets; Monoclonal antibody cocktail; Flow cytometry; HIV disease monitoring; Immunophenotyping) Am J Clin Pathol 1989;92: INFECTION AND SUBSEQUENT DESTRUCTION of CD4 + lymphocytes is a major cause of the immunodeficiency observed in acquired immune deficiency syndrome (AIDS) Decreasing CD4 + lymphocyte levels appear to be the best predictor of morbidity in patients with AIDS 3 and one of the strongest predictors of progression to AIDS in HIV-infected patients. 1 ' 3 ' 1 ' Data demonstrating the significant contribution of several T-cell subsets to the immunopathogenesis of HIV infection have led to inclusion of CD4 and CD8 subset assessments along with functional assays of immune competence in clinical trials of new therapeutic agents. 3 ' " Patients entering therapeutic trials often have advanced disease (i.e., meet Centers for Disease Control [CDC] criteria for AIDS) and significant Received December 28, 1988; received revised manuscript and accepted for publication June, Dr. Liu's current address is Department 367, MS W-3, Beckman Instruments, Inc., 2 S. Kraemer Boulevard, Brea, California Address reprint requests to Dr. Muirhead: Zynaxis Cell Science, Inc., 371 Phoenixville Pike, Malvern, Pennsylvania 193. Department of Immunology, Smith Kline and French Laboratories, Philadelphia, Pennsylvania, and Department of Immunology/Microbiology and OCLS Immunology, Rush- Presbyterian-St. Luke's Medical Center, Chicago, Illinois lymphopenia, and obtaining sufficient blood to support both phenotypic and functional studies is frequently problematic. One approach to reducing sample requirements for immunophenotyping has been described by Horan and colleagues. 6 An antibody cocktail composed of seven different monoclonal antibodies allows simultaneous typing of T-helper (T h ) cells, T-suppressor/cytotoxic cells (T s/c ), B-cells, natural killer () cells, and monocytes in mononuclear cell preparations, using twocolor immunofluorescence and a flow cytometer with a single light source. The technique dilutes commercial fluorochrome-labeled reagents with competing unlabeled reagents to produce discrete fluorescence intensity profiles for each subset of interest. As originally described, 6 the multiple subset cocktail has several possible limitations for immune cell monitoring in HIV-infected patients. First, dimly stained T s/c overlap with T h and may cause overestimation of the latter population. Second, although little interindividual variation in staining intensity was seen in normal subjects, significant intensity shifts as a function of disease state might impair discrimination among subsets in patient specimens. Third, the level of interference from granulocytes was unknown because only mononuclear cell populations were studied. Our objective in the studies described here was to modify the cocktail method to provide a means of lymphocyte subset monitoring that (1) optimized the accuracy of CD4 lymphocyte frequency determinations, (2) applied to small volume and/or lymphopenic specimens from HIV-infected patients, (3) applied to both mononuclear and whole leukocyte cell preparations, and (4) maximized the number of subsets determined per test sample, without increasing the complexity of instrumentation required. 721 on December 217

2 722 LIU ET AL. A.J.C.P. December 1989 We have developed a method that simultaneously evaluates five relevant lymphocyte subsets in a single tube, positively discriminates against contamination of lymphocyte gates by monocytes and unlysed erythrocytes, and is compatible with a variety of cell preparation methods, including automated ones. With the use of the modified cocktail on whole leukocyte preparations, complete immunophenotypes have been obtained on 2-/uL samples from specimens with absolute lymphocyte counts as low as 3 X 1 6 /L. Materials and Methods Peripheral Blood Preparation Methods Human peripheral blood was collected from apparently healthy donors in evacuated tubes containing edetate (EDTA) (Becton-Dickinson #63), acid citrate dextrose solution A (ACD) (Becton-Dickinson #466), or potassium heparin (Becton-Dickinson #641). Specimens exchanged between laboratories were held at ambient temperature in anticoagulant and prepared within 36 hours of collection. For leukocyte-enriched preparations, whole blood was lysed after dextran sedimentation as previously described (lysis method C, Muirhead and associates 1 ). After lysis was complete, leukocytes were washed three times at 4 C with phosphate-buffered saline containing 1% (w/v) bovine serum albumin,.1% (w/v) sodium azide, and.4% (w/v) EDTA (PBAz, ph ) and adjusted to a final cell concentration of 2 X 1 7 cells/ml. Aliquots of 2 X 1 6 cells were pipetted into 96-well V-bottom microliter plates; appropriate antibody reagents and sufficient PBAz for a final volume of 1-12 /ul/well were added; and cells were incubated at 4 C for 3 minutes. Stained cells were washed three times with PBAz and resuspended in the same buffer. Cell viability assessed by propidium iodide exclusion 7 was always greater than 9% and was typically greater than 9%. After viability assessment, cells were fixed by addition of sufficient volume of 4% paraformaldehyde to achieve a final concentration of 1.% (w/v) paraformaldehyde and 2- X 1 6 cells/well, then held at 4 C until analysis was performed. For lysed whole blood preparations, 1-2 nl of whole blood was mixed with the appropriate volume of antibody reagent (see below) and incubated at 4 C for 3 minutes in conical-bottom 12 X 7 mm polypropylene tubes. After staining, erythrocytes were lysed as previously described (lysis method A, Muirhead and associates 1 ). After lysis was complete, leukocytes were washed three times at 4 C with PBAz, viability was assessed, and cells were fixed as described for leukocyte-enriched preparations. For Q-Prep T " preparations, 1 nl whole blood was added to antibody reagent (see below), and the sample was then mixed, incubated, lysed, and fixed by an automated procedure according to manufacturer's instructions (EPICS Division, Coulter Corporation, Hialeah, FL). Mononuclear cells were prepared by Ficoll-Hypaque density gradient centrifugation and stained as described, 2 washed and resuspended in phosphate-buffered saline (PBS) containing 3% (v/v) fetal bovine serum and.1% (w/v) azide, fixed in 1% (w/v) paraformaldehyde, and held at 4 C until analysis was performed. Antibody Reagents Fluorescein isothiocyanate (FITC)-conjugated anti- CD2 (FITC-B,), FITC-conjugated anti-cd14 (FITC- Mo2), and phycoerythrin (PE)-conjugated anti-cd 14 (My4-RD1) were purchased from Coulter Immunology (Hialeah, FL). Unconjugated and PE-conjugated anti- CD8 (UC-Leu-2a, PE-Leu-2a), PE-conjugated anti-cd4 (PE-Leu-3a), unconjugated and FITC-conjugated anti- CD3 (UC-Leu-4, FITC-Leu-4), PE-conjugated anti- CD^ (PE-Leu-11), PE-conjugated anti-cd 19 (PE-Leu- 12), FITC and PE-conjugated anti-cd 14 (FITC- and PE- Leu-M3), FITC- and PE-conjugated mouse IGG, and IgG2a control antibodies with irrelevant specificities, PEconjugated anti-cd6 (PE-Leu-19), and FITC-conjugated anti-cd4 (FITC-HLE) were purchased from Becton-Dickinson Immunocytometry Systems (Mountain View, CA). Simultest T and B Cell or Helper/Suppressor Cell Reagents were also purchased from Becton-Dickinson Immunocytometry Systems. Four or more different lots of each reagent were used during the course of these studies. The cocktail formulations used in these studies are given in Table 1. Staining intensities of commercial fluorochrome-conjugated anti-cd3 and anti-cd8 antibodies were adjusted by dilution with the appropriate unconjugated antibody to give well-resolved staining intensities for T- versus B-cells or CD4 versus CD8 cells. Other antibodies were used at the concentrations recommended by the manufacturer. The primary difference between these formulations and the cocktail originally described by Horan and associates 6 is that the staining intensity of the CD8 population was adjusted to be less intense than that of the CD4 population to avoid overlap of dim CD8 + cells with CD4 + cells. Flow Cytometric Analysis Samples were analyzed on EPICS V, EPICS 73, and EPICS C flow cytometers (Coulter Electronics, Inc., Hialeah, FL), using fluorescence excitation of 2- mw at 488 nm. Right-angle light scatter was collected with the use of a 488-nm longpass dichroic mirror and 488-nm bandpass interference filter. FITC fluorescence was collected with the use of a 6-nm short pass dichroic on December 217

3 Vol. 92 No. 6 ONE TUBE MULTISUBSET MONITORING IN HIV Table 1. Antibody Cocktail Formulations (fil antibody/test*) 723 Cell Type Cluster Antibody (Mg/mL)t CT-lt CT-2 CT-3 CT-4 T-cells (a/stcr + ) CD3 CD3 CD8 CD8 CD4 CD4 CD2 CD19 CD16 CD6 CD14 CD14 CD14 CD14 UC-Leu-4 () FL-Leu-4 (NP) UC-Leu-2a(12.) PE-Leu-2a(12.) UC-Leu-3a (2) PE-Leu-3a(1) FL-B1 (NP) PE-Leu-12 () PE-Leu-llc() PE-Leu-19() T-suppressor/ cytotoxic 2 T-helper/inducer B-cells, granulocytes Monocytes FL-Mo2 (NP) FL-Leu-M3 (NP) PE-Leu-M3 (NP) RD1-My4 (NP) 2 2 NP = not provided by manufacturer. * Manufacturer recommended volume for Becton-Dickinson antibodies 2 ^L/test; manufacturer recommended volume for Coulter antibodies ^L/test. Leu-4 volumes for CT-1 converted from previous manufacturer recommended volume of ^L/test 6 to current recommended volume of 2 jil/test. t See Materials and Methods. X Original cocktail described by Horan. 6 Modified cocktails used for this study. mirror and 3 ± 3 nm bandpass filter. PE fluorescence was collected with the use of 7-nm longpass absorbance or 7 ± 2 nm bandpass filters. Before sample analysis, instrument performance was monitored by running 1- tm microspheres (grade II Fullbright spheres, EPICS Division of Coulter Corporation, Hialeah, FL) under standard conditions for laser power, filters, photomultiplier high voltage, and amplifier gain. Mean intensities and coefficients of variation (CVs) were required to fall within 1% of established standard values for all parameters measured on patient specimens (forward and log 9 degree light scatter, "green" and "red" fluorescence). FluoroTrol-GF fluorescein-stained thymocytes (courtesy of Bob Hoffman, Becton-Dickinson Immunocytometry Systems) were used to reproduce immunofluorescence intensity scales on a daily basis. Instrument high voltage was adjusted to bring the brightest FluoroTrol peak to a predetermined value ± 1 channel. When compensation for low levels of fluorescein fluorescence detected in the phycoerythrin channel was used, the degree of electronic subtraction required was established by (1) adjusting the percent subtraction so that all three FluoroTrol populations (unstained, weakly fluorescein stained, and strongly fluorescein stained) gave equivalent signals in the phycoerythrin channel, or (2) using Calibrite beads (Becton-Dickinson Immunocytometry Systems, Mountain View, CA) according to the manufacturer's instructions. Lymphocyte gates were defined with the use of correlated log 9 degree light scatter (L9) versus forward light scatter (FALS) and then validated with the use of a combination of anti-cd4 and anti-cd 14 staining. Dual-color immunofluorescence distributions were collected with the use of both "tight" and "generous" lymphocyte gates. "Tight" lymphocyte gates included 2% or less CD14 + monocytes. "Generous" lymphocyte gates extended further into the region of overlap between monocytes and large lymphocytes but included no more than 2% CD4 duu granulocytes. The frequency of nonleukocytes (% CD4") and of small monocytes (%CD14 + ) within the defined gates was recorded for each specimen. At least, and typically 1, events fulfilling "tight" lymphocyte gates were collected for each sample. Results Cocktail Versus Single-Color Immunofluorescence Initial studies were done with the use of leukocyte-enriched whole blood preparations in which erythrocytes were lysed before antibody staining (see "Material and Methods"). This method is frequently used in our laboratory because of the convenience afforded by the ability to stain and wash large numbers of specimens in a single 96-well microtiter plate. 1 The multisubset cocktail was first modified by use of undiluted PE-anti-CD4 (PE-Leu-3a) and decreasing PEanti-CD8 (PE-Leu-2a) intensity by addition of sufficient UC-anti-CD8 (UC-Leu-2a) to give a well-resolved T s/c population falling below the T h region in intensity (CT- 1 vs. CT-2, Table 1; Fig. I A). To avoid overlap of reducedintensity T s/c with B-cells, the amount of UC-anti-CD3 (UC-Leu-4) was slightly increased, increasing the separation between T- and B-cells on the green axis. A small amount of PE-anti-CD19 (PE-Leu-12) was added to bring on December 217

4 724 LIU ET AL. B. A.J.C.P. December 1989 ) u c «o vt Q> o 3 «- O c 3 E E LLT ::::<;:':^iiirk'11 / : ::-.: Tu M [ :..:.::::::::+f.:...:.< ^ 1^ ;..:" - M MI ii»ii)imi* Q) k. O) o i i imii \ i I ri^w*f*\ T T log green (FITC) immunofluorescence FIG. 1. Cocktail staining: normal individual and HIV-infected patient. Leukocyte-enriched preparations (see Methods) were stained using CT-2 of Table 1. T h = CD3 + CD + ; T s = CD3 + CD4~; B = CD19 + CD2 + ; = CD3"CD16 + ; M = CD14 + ; results expressed as percentage of total lymphocytes (i.e.. CD14 + cells excluded). A. Normal individual with 4% T h, 24% T s, 9% B, and 14.%. B. HIV-infected individual with 2% T h, 74% T s, 1% B, and 1%. B-cells away from the origin on the red axis and to avoid possible overlap of B-cells with any CD3 + 4~8~ (putative y8 T-cell receptor [TCR + ]) population present, the latter phenotype being present only at low frequency in normal individuals. 9 Monocyte red intensity was increased with the use of RD1-anti-CD 14 (RD1-My4) to ensure that there was no overlap between monocytes and B-cells. Table 2 compares Cocktail 2 staining results with singlecolor immunofluorescence in which each of the cocktail components was used separately in its fluorescein-conjugated form. Paired /-tests showed no significant difference between the two staining methods in 16 healthy adults. Electronic compensation for overlap between fluorochromes in the cocktail allowed use of square analysis regions but was not necessary to resolve lymphocyte subsets. Paired-; analysis in 14 healthy adults showed no difference in subset frequencies obtained by analysis of data taken with or without use of electronic compensation (data not shown). Alternate Markers and Cell Preparation Methods A single-blind study on 12 individuals (3 normal, 9 HIV-infected patients) was carried out between our two laboratories, comparing analysis using Cocktail 2 staining on leukocyte-enriched preparations (authors C-ML, KAM) and two-color immunofluorescence on mononuclear cell preparations (authors SPG, ALL). No significant differences were found with the use of paired-? analysis (data not shown). In general, distinct intensity regions were maintained for all cell populations in the cocktailstained patient samples, and intensity shifts from analysis regions set based on the daily normal control were minimal. However, one patient with a very low lymphocyte Table 2. Modified Cocktail CD4 versus Single -Color Phenotyping in Healthy Adults (n = CD8 CD2 16) CD16 Method SC Method CT Method SC Method CT Method SC Method CT* Method SC Method CT Meant SE Mean (SC-CT) P value SC = single-color immunofluorescence; CT = CT-2 of Table 1. * CD19 + CD2 + (see CT-2 of Table I). f Expressed as % positive lymphocytes (SI conversion: % X.1 = fraction positive lymphocytes). on December 217

5 Vol. 92 No. 6 ONE TUBE MULTISUBSET MONITORING IN HIV 72 count gave the cocktail staining pattern shown in Figure \B. Although T-, B-, and monocyte regions remained well defined, the region became considerably less distinct. The CD 16 marker identified by Leu-llc on cells is also present on granulocytes, although typically at lower densities. This presents little problem in mononuclear cell preparations isolated from freshly drawn specimens. However, whole leukocyte preparations or mononuclear preparations from older specimens contain significant numbers of granulocytes. If some of these granulocytes contaminate the lymphocyte gates because of altered light scatter characteristics or very low lymphocyte counts (perhaps the case in Fig. \B), they could be falsely identified as cells. Therefore, an antibody to an marker not cross-reactive with granulocytes (PE-anti-CD6) was substituted for PE-anti-CD16 (PE-Leu-llc) in Cocktail 3. As expected, granulocytes stained with Cocktail 3 had much lower red intensities than those stained with Cocktail 2 (data not shown). Although intensities varied among patients, lymphocyte staining with anti-cd6 was always dimmer than that with anti-cd 16 for a given person and anti-cd6 intensity decreased steadily from to 48 hours. The decrease was less noticeable but still evident in specimens collected in ACD or preservative-free heparin. Somewhat unexpectedly, high green fluorescence was seen for granulocytes after staining with both Cocktails 2 and 3, suggesting high levels of nonspecific staining. This was true regardless of anticoagulant used. Nonspecific green staining was significantly reduced if cells were stained before erythrocyte lysis (whole blood preparation) rather than after (leukocyte-enriched preparation), suggesting either that exposure to dextran and/or 37 C increased nonspecific staining or that the presence of plasma and erythrocytes decreased it. However, the intensity of monocyte anti-cd 14 staining with RD1-My4 was significantly reduced in whole blood preparations lysed after staining, in some cases resulting in partial overlap with the T h region. The intensity of anti-cd 14 staining with PE-Leu-M3 was much less affected by poststain lysis, and Cocktail 4 was therefore selected as the optimal formulation. Optimized Cocktail Versus Dual-Color Immunofluorescence A second single-blind study was carried out on ten specimens (two normal, eight HIV-infected), comparing effects of time, cocktail composition, and method of specimen preparation. Mononuclear cells were prepared from freshly drawn specimens and analyzed by dual-color immunofluorescence on an EPICS C at Site 1 (SPG, ALL). A second sample from the same specimen was shipped at ambient temperature to Site 2 (C-ML, KAM), where leukocyte-enriched and lysed whole blood samples were analyzed by several staining methods on an EPICS 73. Representative results are given in Table 3. Paired-; analysis showed no significant differences among different preparation methods, between cocktails using PE-anti- CD16 or PE-Leu-19 as the marker, between cocktail and dual-color immunofluorescence analysis, or between laboratories. As shown in Figure 2, the optimized antibody cocktail (Cocktail 4) gave excellent results with all specimen preparation methods examined, including automated lysis using the Coulter Q-Prep. However, in the latter case a wash step or at least resuspension of stained cells in fixative was required. Otherwise, resolution of all populations was precluded by increased sample-streamassociated background fluorescence resulting from increased levels of unbound antibodies in the cocktail reagent as compared with single or dual antibody reagents. "Null" and CD Populations Using Optimized Cocktail In incompletely lysed specimens, the frequency of "null" cells in the cocktail-stained preparations was identical to the frequency of anti-cd4~ cells in the control Table 3. Optimized Cocktail versus Dual-Label Phenotyping in HIV-infected Patient* ID No. 768: 12% lymphocytest 3.8 X 1 9 WBCs/L 462 X 1 6 lymphocytes/l %CD3t %CD4f %Tsf't %CD t % f Preparation Method 2C CTll CT19 2C CTll CT19 2C CTll CT19 2C CTll CT19 2C CTll CTI9 FH(labl) 8 nd nd 3 nd nd nd nd 12 nd nd 1 nd nd DL(lab2) WB"(lab2) nd nd 4 4 nd 74 6 nd 6 6 nd 17 2 FH = mononuclear cells, prepared and stained 6 hours postcollection; DL = dextran-sedimented leukocyte-enriched whole blood, prepared and stained hours postcollection; WB = lysed whole blood, prepared and stained hours postcollection. * One often specimens examined; no significant differences among preparation methods (FH, DL, WB), laboratories, or staining methods (2C, CT11, CT19) by paired-/ analysis. t SI conversion; % X.1 = fraction positive lymphocytes. t T, are given as CD4 8 + for dual-marker staining, as CD3 + 4~ for cocktail staining. 2C = FL-anti-CD3 + PE-anti-CDI9 or PE-anti-CD2; FL-anti-CD4 + PE-anti-CD8; FLanti-CD16 + PE-anti-CD6; FL-anti-CD4 + PE-anti-CD14; CTll = CT-3 of Table 1; CTI9 = CT-4 of Table I. 11 Incomplete lysis: results corrected for erythrocyte contamination of lymphocyte gates based on frequency of null cells (see Results). on December 217

6 j 726 A. LIU ET AL. B. A.J.C.P. December 1989 % : >> * V--X:ilF! " A:^ :.:xxxx::h..- :::::xx..- ) : : : :. : : : «: : ' :'.- : : : : ; :. :... / - ^ \ X:»WSSA<: \ 7 :::.:/ ^ Q,x RBC: ' ^ W^-h:--^- M A. 4-~ null p VTT..TT". I.- ; f-j ViTmi "'Y f 'T 1 n Y mr 'i' i i mi ' f - X ; X<" M a / ^ # \. *\ -!l».=. 1'J.:;-:: :.!.. :::=;,-. i -"<::.:.:*^P::t-- ::: / -4:..:itx:. : :: -: :- "T-"" > J T i : i'» : hti' ::i, f : '1i : ri\iiii i 11 inn/ D. M M.:: / L--7."i ( =::;:ij-:-:v-;. :.<?>, *******.::::: / '/. r l. ; - -,'< " : ":..:"-l+::;:!::: v B : '.r * -* "*«v^ * > *! ' * * * * * V V*»>Z «*> < / Ml* > ^l:;: :;!;i : ;ia«k^:.<r:- ' i^r.^rbc^ I II 11 llll 1 IT I I III! 1 I I Mill' j / <-r / I I T\,.:::: :. :\:" / y. v. ":'. / i 1. - : y ^ : : : A il\x-t-±^;f v : : : ;. : ::;::;.: I ; I I " 't\- :?«:% ::; : = : I L. \i.';v-l&<<<-:r- S ' ****'***** * : : ^ :; : I ~^*' *tti#s??<" log green (FITC) immunofluorescence I 1 1 i 1 Miii 1 1 1'n nun FIG. 2. Optimized cocktail staining on different types of cell preparations. Cells prepared as described were stained using CT-4 of Table 1. T h = CD3 + CD4 + ; T, = CD3 + CD4~; T ou = CD3 + CD4"CDr; B = CD19 + CD2 + ; = CD3"CD6 + ; M = CD14+; RBC = null (unstained). Results expressed as % of total lymphocytes (i.e., null and CD14 + cells excluded). A. Leukocyte-enriched preparation (DL, CT19) from patient 768 (see Table 3 for subset frequencies; T u u = 9%). B. Whole blood preparation (WB, CT19) from patient 768 (see Table 3 for subset frequencies; T nu n = 7%); note significant RBC contamination due to poor lysis. C. Mononuclear preparation from a different patient analyzed in Lab 2 on an EPICS C; T h = 2%, T, = 61%, B = 3%, and = 23%; WBC = 4,/mm 3, 2% lymphocytes. Results from dual-marker staining on same preparation: T h = 3%, T s = 62%, B = 3%, and = 27%. D. Q-Prep preparation from same patient as in panel C; T h = 2%, T s = 6%, B = 12%, and =19%. Elevated B-cell levels seen are typical of this type of preparation (A.L.L. and S.P.G., unpublished data). B on December 217

7 Vol. 92 No. 6 ONE TUBE MULTISUBSET MONITORING IN HIV 727 sample used to validate lymphocyte gates within 2% (data not shown). Therefore, an in-sample correction for contaminating erythrocytes was possible, something not easily achieved with the use of dual-color immunofluorescence. In most specimens, a small population of CD3 + 4"8~ was visible at the origin below the CD and CD regions, ranging in frequency from 3 to 9% (Fig. 3). In many cases, this population also appeared to express slightly higher levels of CD3 than the CD or CD populations. In two specimens tested, the frequency of this population agreed well with the frequency of WT31~ cells, suggesting that these may be CD3 + 7TCR + cells. 14 Cocktail Stability The stability of cocktail reagent was excellent. A 2- test mix of Cocktail 4 was tested and found to give equivalent results for more than a month (data not shown). In theory, it might be necessary to adjust for differences in fluorochrome-protein ratio or antibody titer among different lots of antibody. However, in practice, after initial adjustments for new manufacturer-recommended concentrations between Cocktails 1 and 2, we have found no need to adjust relative amounts of fluorochrome-conjugated and unconjugated antibodies in crossing over between lots of antibody. This also suggests excellent control B 4 :.; ; : ;f jl ma. D -^1 1.., >»!!: \ vm\m^-.i :;;s:*i! l:-... ilr 2 - ;;:. : :.:-;s ; : v. V;:iW;:::i;.. FIG. 3. Identification of CD3 + CD4~CD8~ cells. Mononuclear preparations from two normal individuals were stained using CT-3 of Table l; results are displayed as isometric plots (panels A, C; origin at upper left) or dot plots (panels B, D; origin at lowerright). T h, T s, B, M denned as in Figure 2; = CD3"CD6 + : T h = population l; T s = population 2; B = population 3; = population 4; M = population. A. Individual with <1% WT31" (7 TCR + ) lymphocytes; T nu < 1%. B. Individual with 1% WT31" (y& TCR + ) lymphocytes; T nul, = 17%. of lot-to-lot variation in the commercial antibodies used for these studies, which were carried out with the use of at least four lots of each reagent over a two-year period. Discussion Only CD4 + lymphocytes have been unequivocally established as having clinical utility in management of HIVinfected patients. However, possible involvement of other cell types in the disease process has also been recognized. For example, longitudinal studies of seropositive and high-risk seronegative men suggest that increased levels of CD8 + lymphocytes may be an early indicator of disease progression, 311 and progressive depletion of CD3"CD16 + CD6 + cells in HIV-infected patients has also been described. 1 A large fraction of cells also express low levels of CD8, making it critical to be able to distinguish between CD3 + CD8 + and CD3~CD8 + lymphocytes. 4 ' 1 We have developed a method that allows accumulation of information about these additional lymphocyte subsets simultaneously with clinically relevant data on CD4 + cells. The optimized cocktail staining method described here (Formulation 4, Table 1) has a number of advantages for lymphocyte subset monitoring in HIV-infected patients. It is possible in a single tube to assess frequencies of CD , CD , CD3 + 4~8-, CD19/2 +, and CD6 + lymphocytes. In the same tube, it is also possible to identify monocytes (CD14 + ) and unlysed erythrocytes ("null" cells) within the lymphocyte scatter gates, providing two significant advantages. First, it allows accurate calculation of lymphocyte subset frequencies without assumptions about the constancy of contaminating cell frequencies from tube to tube. Second, it avoids the use of restrictive light scatter gates, which may exclude true positive results (e.g., cells, activated lymphocytes) in the attempt to decrease false positive results (e.g., monocytes). This technique also allows clear distinction between CD T-suppressor cells 4 and CD3~CD16 + or CD3~CD6 + cells, 1 without requiring multilaser systems or complex mathematical analysis of multiple two-antibody combinations. CD3 + 4~8~ cells, of potential interest as mediators of nonmajor histocompatibility-restricted cytotoxicity, 8,9 may also be positively identified. The antibody cocktail described here was found to be easily transferred from one laboratory to another and works equally well with mononuclear cell preparations or lysed whole blood preparations (Fig. 2). When granulocyte-containing preparations are used, some interaction exists between the marker chosen and whether specimens are stained before or after erythrocyte lysis (see "Results"). The optimal method for minimizing interference from nonspecific granulocyte staining was found to be staining of whole blood with an anti-cd6 containing on December 217

8 728 LIU ET AL. A.J.C.P. December 1989 cocktail, followed by erythrocyte lysis. For instruments with lower sensitivity, use of anti-cd 16 (Leu-1 lc) as the marker may be preferable. Although it gives increased granulocyte staining, Leu-1 lc typically gives higher staining intensities than Leu-19 and greater stability of staining intensity over time. One disadvantage of poststain lysis is that the larger sample volumes and tube numbers required for assessing multiple antibody combinations per specimen is time consuming compared with handling leukocyte-enriched preparations stained in 96-well microtiter plates, particularly when dealing with large numbers of specimens. 1 However, with the use of cocktail staining the number of tubes needed to obtain the same information is cut from six (e.g., CD3/CD2, CD4/CD8, CD3/CD8, CD16/CD6, CD4/CD14, isotype control) to two (cocktail, isotype control), significantly decreasing both sample preparation and running times. The optimized cocktail reagent is also compatible with an automated whole blood lysis method (Coulter Q-Prep), providing that a wash step is added to remove excess unbound antibodies. A theoretic limitation of the cocktail staining approach is that intensity differences as well as color differences are used to distinguish subpopulations of cells. Thus, in theory, disease-related shifts in antigen density or loss of a specific epitope could negatively impact resolution of different cell types. Such effects have been seen in leukemic or transplant patients treated with anti-cd3 (ALL, unpublished observations) but not for any of the more than 4 normal or HIV-infected specimens examined to date. Likewise, significant lot-to-lot variations in antibody concentration or fluorochrome-protein ratio could also adversely affect resolution but have not to date been found to be a problem in practice. The cocktail mixture is stable for at least a month after it is made up, making this a practical method even in laboratories that analyze relatively small numbers of specimens per day, although it would clearly be advantageous to have a commercial source of premixed cocktail. In summary, the modified antibody cocktail described here enumerates several lymphocyte subsets undergoing clinical study, in the same tube used to evaluate CD4 + levels for prognosis or therapeutic monitoring in HIVinfected patients. The method also addresses several significant quality assurance issues, providing direct withintube assessment of the most common sources of false positive (monocytes) and false negative (erythrocytes) results, while avoiding false negative results from restrictive light scatter gating. Decreased specimen size, reduced specimen preparation and analysis time, and compatibility with a wide variety of sample preparation methods, coupled with increased information content and data reliability, make this an advantageous method for management of the HIVinfected patient. References 1. Andrieu J-M, Eme D, Venet A, et al. Serum HIV antigen and anti- P24-antibodies in 2 seropositive patients: correlation with CD4 and CD8 subsets. Clin Exp Immunol 1988;73: Bray RA, Gottschalk LR, Landay AL, Gebel HM. Differential surface marker expression in patients with CD16 + lymphoproliferative disorders: in vivo model for differentiation. Hum Immunol 1987;19: Fahey JL, Giorgi J, Martinez-Maza O, Detels R, Mitsuyasu R, Taylor J. Immune pathogenesis of AIDS and related syndromes. In: Gluckman JC, Vilmer E, eds. Acquired immunodeficiency syndrome. Paris: Elsevier, 1987: Gebel HM, Anderson JE, Gottschalk LR, Bray RA. Determination of helper-suppressor T cell ratios. N Engl J Med 1987;316:113.. Ho DD, Kaplan JC. Pathogenesis of human immunodeficiency virus infection and prospects for control. Yale J Biol Med 1987;6: Horan PK, Slezak SS, Poste G. Improved flow cytometric analysis of leukocyte subsets: simultaneous identification offivecell subsets using two-color immunofluorescence. Proc Natl Acad Sci USA 1986;8: Horan PK, Loken ML. A practical guide to the use offlowsystems. In: Van Dilla MA, Dean PN, Laerum OD, Melamed MR. Flow cytometry: instrumentation and data analysis. New York: Academic Press, 198: Lanier LL, Phillips JH, Hackett J Jr, Tutt M, Kumar V. Natural killer cells: definition of a cell type rather than a function. J Immunol 1986;137: Lanier LL, Phillips JH. Evidence for three types of human cytotoxic lymphocyte. Immunology Today 1986;7: Muirhead KA, Wallace PK, Schmitt TC, Frescatore RL, Franco JA, Horan PK. Methodological considerations for implementation of lymphocyte subset analysis in a clinical reference laboratory. Ann NY Acad Sci 1986;468: Polk MD, Fox R, Brookmeyer R, et al. Predictors of the acquired immunodeficiency syndrome developing in a cohort of seropositive homosexual men. N Engl J Med 1987;316: Stahl RE, Friedman-Kien A, Dubin R, Marmor M, Zolla-Pazner S. Immunologic abnormalities in homosexual men: relationship to Kaposi's sarcoma. Am J Med 1982;73: Schroff R, Gottlieb MS, Prince HE, Chai LL, Fahey JL. Immunological studies of homosexual men with immunodeficiency and Kaposi's sarcoma. Clin Immunol Immunopathol 1983,27: van de Griend RJ, Gravekamp C, Bolhuis RLH. Human CD3 + 4"8~ TCRa/3~7 + blood T cells, their lytic potential against tumor cells, and their recognition using WT31 monoclonal antibody. Transplant Proc 1988;2: Vuillier F, Bianco N, Montagnier L, Dighiero G. Selective depletion of low-density CD8 +, CD16 + lymphocytes during HIV infection. AIDS Res Human Retroviruses 1988;4: on December 217