OUP user menu

Usefulness of CD11a and CD18 in Flow Cytometric Immunophenotypic Analysis for Diagnosis of Acute Promyelocytic Leukemia

Yi Zhou MD, PhD, Jeffrey L. Jorgensen MD, PhD, Sa A. Wang MD, Farhad Ravandi MD, Jorge Cortes MD, Hagop M. Kantarjian MD, L. Jeffrey Medeiros MD, Sergej Konoplev MD, PhD
DOI: http://dx.doi.org/10.1309/AJCPQU9R3FSLKFMI 744-750 First published online: 1 November 2012


Acute promyelocytic leukemia (APL) is an aggressive disease that requires prompt diagnosis and therapy. Flow cytometry immunophenotyping can serve as a screening test for APL before the results of cytogenetic or molecular testing for t(15;17)(q22;q21)/PML-RARα are often dimly expressed or absent in APL. We used flow cytometry immunophenotyping with an antibody panel including CD11a and CD18 to assess 36 APL and 33 other AML cases. HLA-DR, CD11a, and CD18 were absent in 81% of APL and 12% of other AML cases (specificity, 88%). By further including combinations of HLA-DR−, CD2+, and either CD11a− or CD18−, we identified 92% of APL cases with 85% specificity. These data compare favorably with the combination of HLA-DR−, CD34−, and CD117+ for APL diagnosis, which had a sensitivity of 64% in this study.

Key Words
  • Acute progranulocytic leukemia
  • Flow cytometry
  • CD11a
  • CD18
  • β2-integrin

Acute promyelocytic leukemia (APL) associated with t(15;17)(q22;q21)/PML-RATα is a clinically aggressive type of acute myeloid leukemia (AML). Affected patients are at risk for developing disseminated intravascular coagulation, and the disease can be lethal if there is a delay in diagnosis and initiation of appropriate therapy. For these reasons, APL is considered a medical emergency because early diagnosis and initiation of therapy are critical to successful patient management. Nevertheless, the overall prognosis of patients with APL is excellent because all-trans retinoic acid and arsenic trioxide combination chemotherapy is highly effective. More than 90% of patients with APL respond to therapy and the 5-year event-free survival is as high as 90%.1,2

Definitive diagnosis of APL requires cytogenetic or molecular evidence of t(15;17)(q22;q21) or the PML-RARα gene fusion. The results of cytogenetic and molecular tests, however, are often not available on the same day the bone marrow (BM) sample is obtained and processed. Therefore, traditionally the initial diagnosis of APL has been established primarily on the basis of morphologic examination. Cases of APL with classic morphologic findings are distinctive—the promyelocytes have hypergranular cytoplasm, often obscuring the nucleus, and multiple Auer rods. It has been long recognized, however, that the morphologic findings in APL are heterogeneous and that a subset of cases has variant features, the so-called microgranular variant.3-5 For this reason, and because flow cytometry immunophenotypic analysis has a short turnaround time, with results often available the same day as the BM testing, flow cytometry immunophenotyping has been used as an ancillary method to establish the diagnosis of APL. Neoplastic promyelocytes, because of their highly granular cytoplasm, have high side scatter. Previous studies using flow cytometry methods have shown a characteristic expression pattern in a large subset of APL cases, which is HLA-DR-, CD34-, and CD117+. This pattern, however, is neither sensitive nor specific for APL.

Some investigators have suggested that the β2-integrin adhesion molecules could be useful for establishing the diagnosis of APL.6,7 β2-integrins are heterodimeric cell adhesion proteins, formed by 1 α chain—CD11a, CD11b, or CD11c—and 1 β chain, CD18. β2-integrin is normally expressed on maturing granulocytes, as well as most types of AML, but these molecules are often not expressed or only dimly expressed by the neoplastic promyelocytes of APL. The diagnostic usefulness of β2-integrin subunits in APL immunophenotypic diagnosis, especially CD11b and CD11c, has been reported in limited studies.8,9 Data on CD11a and CD18 and their potential usefulness in APL diagnosis are also scarce.6,10

The goal of this study was to examine the usefulness of CD11a and CD18 as part of a panel of antibodies used for flow cytometry immunophenotyping of APL cases.

Materials and Methods

Study Group

Following approval by the institutional review board, this retrospective study was performed on cases seen from 2007 to 2009 at the Department of Hematopathology at The University of Texas MD Anderson Cancer Center, Houston. Cases of APL and those non-APL types of AML that were morphologically suspected to be APL were included. The diagnosis in all APL cases was confirmed by the results of conventional cytogenetic analysis, fluorescence in situ hybridization, or polymerase chain reaction (PCR) analysis that showed the presence of t(15;17)(q22;q21) or PML-RARα fusion gene. These methods have been reported previously.11 The non-APL group included AML with morphologic or immunophenotypic features raising a possibility of APL (folded nuclei, granular abundant cytoplasm, and very strong cytochemical myeloperoxidase reaction, and/or absence of CD34 and HLA-DR expression) but subsequently proven to be negative for t(15;17)(q22;q21) by means of karyotype and PML-RARα fusion gene analysis with PCR.

Flow Cytometry Immunophenotypic Analysis

We performed multicolor flow cytometry immunophenotypic analysis on BM aspirate samples collected in EDTA. After incubation of cells with monoclonal antibodies for 10 minutes at 4°C, erythrocytes were lysed with ammonium chloride for 10 minutes, followed by 2 washing steps using phosphate-buffered saline solution. The cells were resuspended and fixed with 1% paraformaldehyde. For this study, all cases were assessed for CD11a (fluorescein isothiocyanate [FITC], clone G25.2) and CD18 (phycoerythrin [PE], clone 6.7). The panel of antibodies used for analyzing AML cases in our laboratory has evolved over the years, and thus the following mouse monoclonal antibodies were applied to at least a subset of cases in this study group: CD45 (peridinin chlorophyll protein [PerCP-Cy5.5-CE], clone 2D1), CD34 (FITC-conjugated, clone 8612), CD117 (PE, clone 104D2), CD33 (FITC or PE, clone P67.6), CD13 (PE, clone L138), CD14 (allophycocyanin [APC], clone MØP9), CD64 (PE, clone 10.1), CD10 (FITC, clone J5), CD19 (FITC, clone 4G7), CD2 (APC, clone L303.1), CD3 (APC, clone SK7), CD5 (APC, clone L17F12), CD7 (APC, clone M-T 701), CD15 (APC, clone H198), CD20 (APC, clone L27), CD38 (APC, clone HB7), CD41 (FITC, clone SZ22), CD56 (FITC, clone NCAM16.2), HLA-DR (FITC, clone L243), and TdT (FITC, 6100). Most antibodies were obtained from BD Bio-sciences (San Jose, CA); anti-CD10 and anti-CD41 were obtained from Beckman Coulter (Brea, CA); and anti-TdT was obtained from Supertechs (Rockville, MD).

The analyses were performed using FACSCaliber cytometers; data were acquired and analyzed using Cell-Quest software (BD Biosciences). For each antibody, negative staining levels were set by comparison with an isotype-matched control. Although a cutoff of 20% is usually accepted as evidence of antigen expression, we found this cutoff suboptimal for APL diagnosis because the distinction between APL and AML of other types was primarily based on absence of expression of certain markers. We therefore tested various cutoffs and found that the optimal combination of sensitivity and specificity was achieved with a cutoff of 30%. We used this cutoff for CD11a, CD18, CD34, and HLA-DR in this study.

Statistical Analysis

The Fisher exact test was applied for categorical variables, and the Student t test was applied for continuous variables. Differences between groups were considered statistically significant if P values were less than .05 in a 2-tailed test.


Study Group

The study group included 69 cases of AML, including 36 APL and 33 non-APL cases. The 36 cases of APL included 6 with a microgranular variant (M3v). Patients in this group included 21 women and 15 men with a median age of 49 years (range, 22-78 years). All patients in the APL group responded to chemotherapy including all-trans retinoic acid and arsenic trioxide treatment.

The non-APL cases included 6 AMLs without maturation, 12 AMLs with maturation, 8 AMLs with multilineage dysplasia, 5 AMLs with monocytic differentiation, 1 AML not further classified, and 1 AML morphologically closely resembling APL but associated with t(11;17)(q23;q21). Among these cases, 19 were men and 14 were women, with a median age of 65 years (range, 13-83 years). These patients, except one with AML associated with t(11;17)(q23;q21), were treated with standard chemotherapy regimens without all-trans retinoic acid or arsenic trioxide.

Flow Cytometry Immunophenotypic Results

Analysis of side scatter vs CD45 expression showed that APL cells with typical hypergranular morphology displayed high side scatter. In contrast, APL cells with microgranular morphology showed intermediate to high side scatter, but less than that seen in cells with typical morphology. The blasts of non-APL cases showed substantially lower side scatter.

The percentages of positive cells in a blast-promyelocyte gate for diagnostic markers used in flow cytometry immunophenotyping for APL and AML are shown in Figure 1A and Figure 1B, respectively. Applying a cutoff of 20% or 30%, a profile of positivity of those markers for cases of APL and AML other than APL is shown in Figure 2.

Figure 1

Expression of immunophenotypic markers in acute promyelocytic leukemia (A) and acute myeloid leukemia of other types (B). The y- axis represents percentage of positive cells in blast-promyelocyte gate.

Figure 2

Flow cytometric immunophenotypic profile of acute myeloid leukemia. Rows represent each case and columns represent each flow cytometric immunophenotyping marker. Red, yellow, and white color indicate that more than 30%, 20% to 30%, and fewer than 20%, respectively, of cells in blast-promyelocyte gate are positive for each marker.

With a cutoff of 20% for CD2, CD13, CD15, CD33, CD64, and CD117, and 30% for CD11a, CD18, CD34, and HLA-DR, most APL cases were positive for CD33 (36/36; 100%), CD13 (35/36; 97%), CD117 (30/36; 83%), and CD64 (29/36; 81%). A subset of APL cases was positive for CD15 (23/36; 64%) and CD2 (12/36; 33%). Most APL cases were negative for HLA-DR (35/36; 97%), CD11a (33/36; 92%), CD18 (33/36; 92%), and CD34 (30/36; 83%). The microgranular variant of APL showed significantly higher frequency of CD2 (5/6) than the classic variant (8/31; P < .05).

In comparison, most cases of AML of other types expressed CD33 (32/33; 97%), CD13 (30/33; 91%), and CD117 (30/33; 91%). A substantial subset of these cases was negative for CD34 (19/33; 58%), HLA-DR (15/33; 45%), CD11a (13/33; 39%), or CD18 (15/33; 45%). The differences in frequency of expression of CD34, HLA-DR, CD11a, CD18, and CD2 between APL and other types of AML were statistically significant (P < .05).

Absence of either CD11a or CD18 was observed in all 36 (100%) cases of APL. These results compared well with the absence of HLA-DR, the most sensitive single marker. All other single markers had lower sensitivity. Single marker analysis, however, lacked specificity, because CD11a, CD18, or HLA-DR was absent in 39%, 45%, and 45% of non-APL cases, respectively.

Various combinations of markers including CD11a, CD18, and HLA-DR were assessed to best distinguish APL from non-APL cases Table 1. The combination of HLA-DR− and either CD11a− or CD18− was very frequent in APL, as observed in 35 (97%) of 36 APL cases in this study. However, 13 (39%) of 33 non-APL cases also showed these combinations of findings. The combined absence of CD11a, CD18, and HLA-DR was observed in 29 (81%) of 36 APL cases and in 4 (12%) of 33 non-APL cases (P < .05).

View this table:
Table 1

The combination of CD2+ and HLA-DR− was also highly associated with APL (P < .01). All 12 CD2+ APL cases were negative for HLA-DR expression, as defined by a cutoff of 30%. Furthermore, CD2 expression was highly associated with APL cases in which the immunophenotype was atypical. Among 11 APL cases that were positive for CD34, CD11a, or CD18, 8 cases (73%) were CD2+, whereas only 4 (8%) of 22 APL cases with a typical immunophenotype were CD2+ (P < .01). In contrast, CD2+ was uncommon in non-APL AML cases and usually associated with HLA-DR expression. We therefore included CD2 in antibody combinations for potential APL diagnosis. In APL, 33 (92%) of 36 cases were either triple negative for CD11a, CD18, and HLA-DR; or HLA-DR−, CD2+, and CD11a−; or CD18− Figure 3. These patterns were present in only 5 (15%) of 33 cases of AML of non-APL type, with a specificity at 85% (P < .01).

Figure 3

An example of HLA-DR−, CD2+, CD34+, CD11a−, and CD18−acute promyelocytic leukemia associated with t(15;17) (q22;q21).

Earlier studies have shown the value of the combination of HLA-DR−, CD34−, and CD117+ for the diagnosis of APL. In this study, the combination of these markers was observed in 23 (64%) of 36 of APL and 11 (33%) of 33 of non-APL AML, with a specificity at 67%.


Others have shown the value of flow cytometry immunophenotyping as an ancillary test for the diagnosis of APL. Various antibody combinations have been proposed, with perhaps the best known combination being HLA-DR−, CD34−, and CD117+. More recently, CD11a and CD18 have been suggested as being helpful for establishing the diagnosis of APL.6,10 However, relatively few studies have been performed on large numbers of APL and non-APL AML cases, nor have antibody combinations been rigorously compared in the same study group.

In this study, we show that the combined absence of CD11a and CD18 is a useful immunophenotypic finding that supports the diagnosis of APL. In approximately 83% of APL cases, CD11a and CD18 were not expressed. However, a small subset of APL cases, approximately 8%, showed expression of CD11a or CD18 in more than 30% promyelocytes. The sensitivity of the combination of CD11a− and CD18− was 83%, with a specificity of 79%, in comparison with a set of APL-like AML cases. In this study, CD11a and CD18 were negative in 1 HLA-DR+ APL case.

The skewed distribution of CD11a and CD18 expression in APL is in keeping, in large part, with the results of earlier studies. Di Noto et al12 showed that CD11a, CD11b, and CD11c were not expressed (defined as <20% of events being negative) in 93%, 94%, and 89% of APL cases, respectively. Similarly, Paietta and colleagues6 also reported that 75% of APL cases showed CD11a expression in fewer than 10% of cells, though a small subset of APL cases expressed CD11a above the 20% cutoff. Other studies, however, showed that absence of CD11b and CD11c occurs in nearly all APL cases.8,9 These discrepancies, although minor in degree, suggest that expression levels of β2-integrin subunits differ independently in APL cells. It also has been shown that expression levels of CD11a, CD11b, and CD11c respond differently to all-trans retinoic acid treatment, with CD11a expression being the least responsive.13 Considering these earlier stud-ies,8,9 it appears that CD11b and CD11c are more consistently absent in APL than CD11a and CD18. However, variation of technical factors, such as antibody source, titer, fluorochrome combinations, and thresholds established for positivity can contribute to these apparent discrepancies in assay sensitivity.

It is well known that CD2, another cell adhesion molecule, is expressed in a subset of APL cases, mostly the microgranular variant.14,15 CD2 expression is less often used as a diagnostic marker for APL, however, because CD2 expression is considered nonspecific. CD2 is expressed in AML associated with inv(16)(p13.1q22)16 and in mast cell disease17 in addition to APL. Here, we show that assessment of CD2 has complementary value to other markers in the immunophenotypic diagnosis of APL. In the absence of HLA-DR, CD2 expression is highly associated with APL, especially with the subset of cases positive for CD34, CD11a, or CD18. In fact, 5 of 8 CD2+ APL cases were of the microgranular variant; conversely, 5 of 6 microgranular variant cases were CD2+. In this study, the combinations of HLA-DR−, CD11a−, and CD18−, and HLA-DR−, CD2+, and either CD11a− or CD18− identified 92% of APL cases, with 85% specificity. It also should be noted that the specificity described in this study is based on a set of non-APL AML cases that had at least some features suspicious for APL based on clinical, morphologic, and/or immunophenotypic findings. This subset of cases, therefore, may not represent the true frequency of CD11a and CD18 expression in non-APL AML cases. However, the non-APL cases in this study represent a subset of AMLs most likely to be misdiagnosed as APL, and thus these findings emphasize the usefulness of CD11a and CD18 in identifying true APL cases.

Others have combined the results of various antibody stains in an attempt to improve the sensitivity and specificity of APL diagnosis with flow cytometry immunophenotyping. Perhaps the best-known combination to support the diagnosis of APL in the literature is HLA-DR−, CD34−, and CD117+. In the current study, this combination of results was 64% sensitive and 67% specific. In contrast, the combination of HLA-DR−, CD11a−, and CD18− was 81% sensitive and 88% specific. We therefore conclude that the inclusion of CD11a and CD18 in various antibody combinations is a superior approach for immunophenotypic diagnosis of APL. The inclusion of CD2 in an antibody panel was a further improvement. The combination of HLA-DR−, CD11a−, and CD18−, and HLA-DR−, CD2+, and either CD11a or CD18− had 92% sensitivity and 85% specificity.

In summary, we assessed the usefulness of including the β2-integrin subunits CD11a and CD18 in an antibody panel for diagnosis of APL using flow cytometry immunopheno-typic analysis. We found CD11a and CD18 to be absent or only dimly expressed in most APL cases. After assessing various antibody combinations, we believe the combination of HLA-DR−, CD2+, and either CD11a− or CD18− constitutes an immunophenotype that reliably supports the diagnosis of APL.


View Abstract