OUP user menu

Minimal Residual Disease Detection by Flow Cytometry in Adult T-Cell Leukemia/Lymphoma

Haipeng Shao MD, PhD, Constance M. Yuan MD, PhD, Liqiang Xi MD, Mark Raffeld MD, John C. Morris MD, John E. Janik MD, Maryalice Stetler-Stevenson MD, PhD
DOI: http://dx.doi.org/10.1309/AJCPS1K0OHLJYWWV 592-601 First published online: 1 April 2010


Little information exists regarding the detection of minimal residual disease (MRD) in adult T-cell leukemia/lymphoma (ATLL). We evaluated 75 peripheral blood samples from 17 ATLL cases using flow cytometry (FC); 50 of the samples were concurrently evaluated by polymerase chain reaction (PCR) for clonal T-cell receptor γ chain (TRG) gene rearrangement and the presence of human T-cell lymphotropic virus-1 proviral sequences. Residual ATLL cells were identified using a multiparametric approach to identify aberrant T-cell immunophenotypes. Malignant T cells were CD4+, CD3 dim+, CD26–, CD25 bright, CD7+, and CD27+, with occasional dim expression of CD2 or CD5. FC exhibited a high sensitivity, detecting as few as 0.29% ATLL cells/WBC (4.9 cells/μL) in the peripheral blood. PCR for TRG gene rearrangement was slightly more sensitive, and FC and PCR complemented each other in detecting MRD. In 2 patients, there was complete remission; 4 patients had disease refractory to therapy, and 3 died; 11 others had persistent disease with variable numbers of ATLL cells in the peripheral blood. Higher levels of ATLL cells appeared to correlate with disease severity. FC detection of aberrant T cells permits sensitive and quantitative monitoring of MRD in ATLL.

Key Words:
  • Adult T-cell leukemia/lymphoma
  • Minimal residual disease
  • Flow cytometry
  • Immunophenotype
  • CD26
  • Human T-cell lymphotropic virus-1
  • HTLV-1
  • Polymerase chain reaction

Adult T-cell leukemia/lymphoma (ATLL) is a peripheral T-cell neoplasm caused by human T-cell lymphotropic virus-1 (HTLV-1).13 ATLL is most frequently seen in HTLV-1–endemic areas such as the Caribbean basin, southern Japan, central and south Africa, and South America. HTLV-1 infects approximately 20 million people worldwide, with ATLL subsequently developing in 2% to 4% of that population.35 Cases of ATLL in North America are usually found in immigrants from endemic areas.6 ATLL is diagnosed based on characteristic clinicopathologic features and the serologic presence of HTLV-1 infection.

Most patients with ATLL have systemic disease involving lymph nodes, peripheral blood, spleen, and extranodal sites such as skin.7 Based on clinical manifestations and laboratory findings, ATLL is classified into 4 subtypes: acute, lymphomatous, chronic, and smoldering.8 The majority of ATLL cases are the aggressive acute and lymphomatous subtypes with a poor prognosis and median survivals of 6.2 months and 10.2 months, respectively.8 The rarer chronic and smoldering ATLL subtypes are indolent and usually managed with expectant observation. The treatment for aggressive ATLL is usually combination chemotherapy using cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)-like regimens or a combination of zidovudine and interferon alfa.9,10 The overall survival is still poor due to frequent and early disease relapses.

Because peripheral blood is most consistently involved in ATLL, detection of ATLL cells in the peripheral blood is reliable in monitoring the minimal residual disease (MRD) after treatment. The so-called flower cells with lobulated nuclei in the peripheral blood are characteristic of ATLL; however, ATLL cells may show a broad cytologic appearance.11 In addition, the low level of disease involvement in MRD is usually below the limit of detection by morphologic studies.

Flow cytometry (FC) has been shown to be highly sensitive in detecting MRD in acute myeloid and acute lymphoblastic leukemia. Furthermore, FC offers more accurate quantitation of MRD levels.1214 The ATLL cells exhibit a characteristic immunophenotype with expression of CD3, CD2, CD4, CD5, and bright CD25 and loss of CD7, although occasional cases are double positive for CD4 and CD8 and may have partial or dim CD7 expression. In 1 study, approximately half of the ATLL cases showed dim CD3 expression, allowing better distinction from normal T-cell populations.15

In monitoring MRD by FC, it is often difficult or impossible to completely separate ATLL cells from normal T-cell populations in the peripheral blood using any single set of markers. Normal CD7– T cells may express CD4 and T-cell receptor (TCR) α/β proteins and often show lower levels of expression of CD3.16,17 Hence, additional markers are needed to permit more definitive identification of ATLL cells in MRD. CD26 is a surface glycoprotein expressed predominantly by peripheral blood CD4+ T cells.18,19 Loss of CD26 has been observed in cutaneous T-cell lymphomas involving the peripheral blood20,21 and ATLL.22,23 Therefore, addition of CD26 to the standard T-cell panel has the potential for improved detection and more precise quantification of residual ATLL cells.

To our knowledge, no previous reports exist that describe the monitoring of MRD by FC in ATLL. In this study, serial peripheral blood samples from 17 patients with ATLL undergoing treatment were evaluated by FC for MRD based on analysis of CD26–, CD4+, CD7–, CD25 bright+, and CD3 dim+ T-cell subpopulations. The sensitivity of routine FC was determined and compared with that of polymerase chain reaction (PCR) for T-cell receptor γ chain (TRG) gene rearrangements and quantitative PCR for HTLV-1 proviral sequences. The levels of MRD were monitored sequentially and correlated with clinical findings.

Materials and Methods

Patient Information

For the study, 17 patients with the diagnosis of ATLL were evaluated by FC before and after treatment; 8 patients were men, and 9 were women. Their median age was 50 years (range, 15–75 years). The patients had newly diagnosed or relapsed ATLL and were enrolled in treatment protocols evaluating investigational agents at the National Cancer Institute, Bethesda, MD. All ATLL diagnoses were confirmed by pathologic review and/or immunophenotypic analysis of the peripheral blood. Clinical information was obtained through chart review. The patients were treated with single-agent daclizumab, alemtuzumab, or denileukin diftitox or with multiagent chemotherapy regimens (CHOP or etoposide, vincristine, doxorubicin, cyclophosphamide, and prednisone [or EPOCH]) alone or in combination with the aforementioned agents. In the study, 75 peripheral blood samples were obtained from the 17 patients and analyzed by FC for MRD. Concurrent molecular analyses for TRG gene rearrangement and HTLV-1 proviral sequences were performed in 50 peripheral blood samples. Bone marrow biopsy specimens were evaluated by morphologic studies and immunoperoxidase stains for tumor cells. Concurrent CBCs with differential counts were obtained with all 75 samples.

FC Studies

All specimens were stained within 24 hours of collection with a panel of antibodies that included CD2-phycoerythrin (PE), CD3-peridinin chlorophyll protein (PerCP), CD4-fluorescein isothiocyanate (FITC), CD4-PerCP, CD5-allophycocyanin (APC), CD5-FITC, CD8-PE, CD10-APC, CD14-FITC, CD16-PE, CD19-PerCP, CD25-PE, CD26-FITC, CD27-PE, CD34-APC, CD38-PE, CD45-PerCP, CD56-PE, TCR α/β-FITC, and TCR γ/δ-PE (BD Biosciences, San Jose, CA); CD3-APC and CD4-APC (IMMUNOTECH, Marseilles, France); CD7-FITC and CD52-PE (SouthernBiotech, Birmingham, AL); and CD13-PE (Beckman Coulter, Brea, CA). Briefly, RBCs were lysed by incubating with lysing solution (150 mmol/L NH4Cl, 10 mmol/L KHCO3, 0.1 mmol/L EDTA) at a ratio of 1:9 (volume of sample/volume of lysing solution) for 10 minutes at room temperature. The specimens were washed with phosphate-buffered saline (PBS) to remove cytophilic antibodies before determining cell number. Cell counts were determined using a hemocytometer, and viability was determined by trypan blue exclusion. The specimens were stained at room temperature for 30 minutes with a cocktail of 4 antibodies at antibody concentrations according to the manufacturer’s recommendations. After incubation, cells were pelleted by centrifugation (500g for 15 minutes at room temperature), media was aspirated, and the cells were washed twice in 0.1% NaN3 in PBS. The cells were fixed in 1% formalin in PBS and stored at 4°C for up to 12 hours before FC acquisition.

We performed 6-parameter, 4-color FC with a BD Biosciences FACSCalibur flow cytometer (BD Biosciences). The sensitivity of fluorescent detectors was set and monitored using FACSComp software using Calibrite beads (BD Biosciences) according to the manufacturer’s recommendations. Data, collected in list mode, were analyzed with CellQuest Pro software (BD Biosciences) and FCSExpress (De Novo Software, Los Angeles, CA). At least 5,000 lymphocytes were acquired per tube. For analysis, lymphocytes, monocytes, and granulocytes were gated by forward vs side scatter. CD19+, CD3+, and CD14+ cells were back-gated to determine the appropriateness of the gates. Analysis gates based on CD3 positivity were also used to study antigen expression in T cells. Normal lymphoid cells within the specimens served as internal controls for determination of antibody binding intensity. The number of malignant cells per microliter was determined as previously described.24 The percentage of tumor cells was determined in light scatter–based lymphocyte gates and applied to the absolute lymphocyte count or absolute lymphocyte plus atypical cell counts as derived from the concurrent CBC data.

PCR Studies of TRG Gene Rearrangements and HTLV-1

Genomic DNA was extracted from citrate-anticoagulated whole blood using the automated Magtration System 8 Lx (PSS Precision System Science USA, Livermore, CA) according to the manufacturer’s instructions. To determine clonality of the TRG locus, PCR was performed on each sample using 3 sets of primer pairs. Set I PCR with primers Vγ101, Vγ11, Jγ1, and Jγ2 and set II PCR with primers Vγ101, Vγ11, Jp1, and Jp2 were as described by McCarthy et al.25 Set III primers Vγ9-150 (5′-CGT CTA CAT CCA CTC TCA C-3′), Jγ1, and Jγ2 (5′-CAA GTG TTG TTC CAC TGC C-3′) were developed and modified in house. Positive control DNA for the set I, II, and III reactions were from the CEM, MOLT-4, and HSB cell lines, respectively. For the PCR reactions, 1 μg of DNA template was mixed with 1× PCR buffer containing 2.0 mmol/L magnesium chloride, 2.5 mmol/L deoxynucleoside triphosphates, 0.3 μmol/L of each primer (sets I and II) or 0.5 μmol/L of each primer (set III), and 2.5 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA). The DNA was amplified in a thermocycler (GeneAmp PCR System 9700, Applied Biosystems) for 4 minutes at 94°C followed by 35 cycles (94°C, 1 minute; 60°C, 1 minute; and 72°C, 1 minute) with a final cycle of extension at 72°C for 7 minutes. After amplification, the products were analyzed by 16% nondenaturing polyacrylamide gel electrophoresis and visualized by UV light after ethidium bromide staining.

For detection of HTLV-1 in peripheral blood mono-nuclear cells, PCR amplification for HTLV-1 pol gene was performed using TaqMan technology on the Applied Biosystems Model 7900. Consensus primers SK110 and SK111 (Sigma-Genosys, St Louis, MO) for the pol region of the proviral DNA and an internal TaqMan fluorogenic probe SK112 for HTLV-1 (FAM-CTT TAC TGA CAA ACC C-MGB) were modified from probes published by Segurado et al.26 Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous control gene target. The 50-μL PCR mixture consisted of standard templates or 1 μg of sample DNA extract, primers SK110 and SK111 (500 nmol/L of each), and 100 nmol/L of HTLV-1 probe, in 1× Universal Master Mix Buffer with AmpErase UNG (Applied Biosystems). The PCR conditions were as follows: 2 minutes at 50°C (to activate AmpErase reagents), denaturation for 10 minutes at 95°C, followed by 40 amplification cycles (95°C, 15 seconds; and 60°C, 60 seconds). All standard dilutions, controls, and patient samples were run in duplicate, and the average value of the copy number was used to quantify HTLV-1.

A test result was valid when the slope was −3.5 ± 0.2 (corresponding to PCR efficiencies of between 86% and 100%); the coefficients of correlation, r 2, were more than 0.98; positive controls were positive and fell in range; cycle thresholds of negative control (commercial placental DNA) and no template control samples were 40 or more cycles; and the coefficient of variation of virus copies in duplicate runs for each sample was 20% or less. If nonvalid results occurred, the test was repeated. The HUT cell line infected with HTLV-1 was used as the quantitation control, based on a reported infectivity of 17 proviral copies per cell.27 An acceptable quantitation range was considered to be 255 to 255,000 copies per microgram DNA based on the standard curve. The results were considered negative when the HTLV-1 cycle threshold was 40 cycles or more.


We evaluated 75 peripheral blood specimens from 17 patients with the diagnosis of ATLL by FC. The immunophenotypic data are summarized in Table 1 and Table 2. ATLL cells were consistently positive for CD4, CD2, CD25 (bright), CD27, and TCR α/β and negative for CD26 and TCR γ/δ. CD3 was dimly positive or negative in all samples. CD7 was lost, dimly expressed, or partially expressed in all but 2 samples. About 22% of cases also showed loss or dim expression of CD5. In most patients (14/17), the immunophenotype of the ATLL cells was consistent among all samples from the same patient despite treatment. However, significant differences in the apparent levels of CD3 expression (shifting between normal and dimly positive) were observed in sequential samples from patients 9 and 11, and CD7 expression normalized in some samples from patients 11 and 13. With a high tumor load, the leukemic cell population can be reliably identified as CD3 dim/CD4+/CD7–/CD25+ Image 1A. After treatment, responsive patients showed a marked reduction of leukemic cells Table 3, making detection more challenging. Residual ATLL cells can be separated from the normal CD7– T cells by using a multiparametric approach identifying CD26 negativity in CD4+ T cells Image 1C, dim CD3 expression, and high CD25 expression. CD26 negativity is particularly useful when the expression levels of CD3 and/or CD7 are near normal Image 1B (Image 1C). With this strategy, as few as 0.29% ATLL cells/WBC (4.9 cells/μL) can be identified by FC.

View this table:
Table 1

Of the 75 samples, 50 concurrent PCR assays for TRG rearrangement and 48 PCR assays for HTLV-1 viral DNA were also performed. Clonal TRG rearrangements were demonstrated in 37 samples (74%), and an additional 7 samples (14%) were “suspicious” for a clonal TRG rearrangement. The clonal bands were identical in size in the same patient, except 1 specimen each in patients 1 and 10. HTLV-1 proviral DNA sequences were detected in 48 samples (96%). Of the 50 samples, FC and PCR for TRG rearrangement identified MRD in 30 (60%) of 50 samples. FC and PCR for TRG rearrangement complemented each other in establishing MRD when one method had results that were suspicious but not diagnostic for ATLL in 6 samples (12%). Both methods were indeterminate in 2 samples (4%), and both were negative in 4 samples (8%). FC detected ATLL cells in 2 samples (4%) in which PCR failed to show clonal TRG rearrangement. PCR detected clonality in 4 samples (8%) and was indeterminate in 2 samples (4%) in which FC results were negative. PCR for TRG rearrangement was slightly more sensitive than FC and can detect at least 0.09% ATLL cells/WBC (2.5 cells/μL), a result considered suspicious by FC.

Fifteen bone marrow biopsies were performed in 11 patients; 8 biopsy specimens from 7 patients showed an atypical T-cell infiltrate consistent with involvement by ATLL, and FC of the concurrent peripheral blood samples showed circulating ATLL cells in all. Five bone marrow biopsy specimens from 4 patients were negative for involvement by ATLL; however, the corresponding peripheral blood showed MRD by FC. Patient 16 achieved complete remission, as evidenced by negative bone marrow biopsy, FC, and PCR results.

The levels of ATLL involvement were monitored in sequential peripheral blood samples from the 17 patients Figure 1. The number of tumor cells per microliter of blood was determined by routine FC. Two patients (3 and 16) responded to therapy with no detectable ATLL cells in the blood by FC, although one of them still had a clonal TRG rearrangement shown by PCR. Both patients went into complete remission as defined by 2 or more negative FC results at least 4 weeks apart. Three patients (4, 7, and 8) died due to progressive ATLL; patient 8 had disease refractory to treatment with persistently elevated levels of ATLL cells in the blood, and patients 4 and 7 responded initially with a marked reduction of ATLL in the blood but quickly experienced relapse. Patient 10 showed a partial response, but overall exhibited persistently high levels of ATLL cells in the blood (>10,000/μL). The ATLL cell levels in patients 4, 7, 8, and 10 after treatment ranged from 29.9 to 134,064/μL with a median value of 3,609.9/μL. The majority of patients (11/17) responded to therapy and went into clinical remission, but the peripheral blood samples showed persistent MRD by FC and PCR. The ATLL cell levels in these patients ranged from 0.66 to 2950.7/μL with a median value of 153.8 cells/μL. Patients 10 and 11 had a significant increase in ATLL levels after an initial response, corresponding to clinical disease progression.

View this table:
Table 2


Detection of MRD by FC has proven to be important for predicting relapse and guiding treatment in acute myeloid leukemia, acute lymphoblastic leukemia, and chronic lymphocytic leukemia12,28,29 and is now in routine clinical use in these diseases. In contrast, there is little information available regarding MRD monitoring of mature T-cell neoplasia, specifically ATLL. Several case reports described PCR-based molecular detection of HTLV-1 or TRG rearrangement to monitor MRD after treatment.30,31 Because ATLL cells exhibit a characteristic immunophenotype (CD3+/CD4+/CD7–/CD25+), FC seems to be well suited for MRD detection. However, CD7– and CD25 bright expression can be observed in normal CD4+ T cells.17,3234 Therefore, separation of ATLL cells from normal CD7–/CD4+ and CD4+/CD25+ T-cell populations requires additional markers.

Yokote et al15 demonstrated abnormally dim CD3 expression in ATLL cells that can be used to separate ATLL cells from normal T cells, but only about half of the samples showed dim CD3 expression in their study. Therefore, more specific markers are needed to better define residual ATLL cells. Multiple studies have shown that CD26 expression is lost on ATLL cells and on cutaneous T-cell lymphoma cells in peripheral blood.20,22,35 In healthy people, CD26 is expressed predominantly on CD4+ T cells (normally greater than 60% of CD4+ T cells are CD26+), and there is a small subset of T cells that are CD4+/CD26–. In the present article, we report on a multiparametric approach to detect malignant T cells by identification of CD3 dim, CD4+, CD7 dim/–, CD25 bright. and CD26– T cells.

The FC findings were correlated with concurrent PCR analysis for TRG rearrangement and HTLV-1 proviral DNA detection. With the multiparametric approach, routine FC can detect minimal residual ATLL cells at levels down to 0.29% (4.9 cells/μL) in the blood, even in the setting of lymphopenia. In contrast with the findings of Yokote et al,15 all of our ATLL cases showed dim or negative CD3 expression (100%) at initial evaluation. The degree of CD3 aberrancy varied from case to case, with some cases having greatly decreased CD3 expression and other cases demonstrating only mildly decreased expression. In all specimens, examination of the expression of CD3, CD7, CD25, and CD26 in CD4+ cells allowed the diagnosis of ATLL. FC and PCR for TRG rearrangements showed high correlation and complemented each other in detecting MRD in ATLL. PCR for TRG rearrangements was slightly more sensitive than FC (0.09% vs 0.29%). It is possible that even greater PCR sensitivity could be obtained by first sorting by FC for CD3+, CD4+, CD7–, and CD26– cells and then performing PCR. The discrepancy between FC and PCR in the remaining samples is unlikely related to the absolute numbers of residual ATLL cells in the blood because some samples with more ATLL cells detected by FC were negative by PCR and vice versa. The positive FC–negative PCR cases may be due to the inherent low false-negative rate (5%–7%) for detection of TRG rearrangements by the PCR method used.

Image 1

Representative flow cytometric analysis of adult T-cell leukemia/lymphoma cells in peripheral blood samples. The lymphoid cells were gated by forward light scatter vs side light scatter. The leukemic cells are indicated in red. A, The leukemic cells show typical immunophenotype: CD3 dim/CD7–/CD4+/CD25+/CD26–. B, The leukemic cells displayed a nearly normal level of expression of CD3 with a small subset dim for CD3 and were CD4+ and CD7–/CD26–. C, The leukemic cells were partially positive for CD7, slightly dim for CD3, and CD4+/CD26–. APC, allophycocyanin; FITC, fluorescein isothiocyanate; PE, phycoerythrin; PerCP, peridinin chlorophyll protein.

In 2 patients (1 and 10), clonal TRG rearrangements distinct from the original diagnostic samples were detected by PCR right after treatment with alemtuzumab. Both corresponded with severe reduction of leukemic cells identified by FC (4.9 cells/μL in patient 1, and no leukemic cells in patient 10). In patient 1, PCR analysis of subsequent samples showed clonal TRG rearrangements identical to the original diagnostic samples. Most likely, these unrelated clones represented restricted T-cell populations during the repopulation process after T-cell depletion therapy with alemtuzumab. In patient 1, the true leukemic cell clone may have been masked by the reactive clone because FC showed a predominance of T cells in the lymphoid cells (96%), and only 12% of the lymphoid cells were leukemic cells.

Figure 1

Minimal residual disease (CD26–/CD4+ cells) in 17 patients with adult T-cell leukemia/lymphoma (ATLL). The time shown on the right is the follow-up period. Patients 4, 7, and 8 were unresponsive to treatment and died of disease progression. Patient 15 continued with daclizumab treatment after initial treatment with denileukin diftitox.

It is interesting to note that HTLV-1 viral DNA was detected by quantitative PCR analysis in almost every sample tested. It is unlikely that positive PCR for HTLV-1 viral DNA in samples negative by both FC and PCR for TRG rearrangement indicates MRD in the blood. This most likely represents a residual carrier state, as once HTLV-1 integrates its provirus into the host T-cell genome, it is very difficult to clear the virus from the body. It is well known that proviral DNA can be detected by quantitative PCR analysis in HTLV-1–infected carriers without ATLL.36 HTLV-1+ cells have also been found to remain after achieving complete remission in patients with ATLL treated with stem cell transplantation.37 Therefore, the presence of HTLV-1 should not be used as an indication of MRD.

Of the 17 patients examined, 4 had disease that was refractory to therapy and had persistent high levels of circulating ATLL cells; 2 patients were shown to be disease free by FC; and 11 patients responded to therapy but showed persistent MRD in the blood. The levels of MRD in the 11 patients ranged from several to hundreds of ATLL cells per microliter (range, 0.66–2,950.7/μL; median, 153.8/μL). Four patients subsequently showed an increase of ATLL cells to thousands of cells per microliter that corresponded to clinical disease progression (range, 29.9–134,064/μL; median, 3,609.9/μL). Given that few patients with MRD demonstrated disease progression during the period studied, the statistical significance of the different levels of MRD with disease progression or relapse cannot be confirmed in this study. Future studies involving a larger group of patients with disease progression after initial response to treatment are necessary to establish the level of MRD associated with outcome.

Our study confirms the finding of down-regulation or complete loss of CD26 in ATLL cells reported by other groups.22,23 The loss of CD26 expression has been observed in other T-cell lymphoma/leukemias, including Sézary syndrome (SS), CD4+ peripheral T-cell lymphoma, T-cell large granular lymphocytic leukemia, and T-cell hepatosplenic lymphoma.20 Although more recent data suggest that the CD26 marker has lower sensitivity (41.1%–63.6% positive) and specificity for SS,21 the multiparametric approach used in this study, namely examining CD3 dim (frequent in SS), CD4+, CD7–, CD25 homogeneously positive or negative, and CD26–, may increase the sensitivity. Given the almost universal loss of CD26 in our ATLL cases, CD26 negativity has a high sensitivity for detecting ATLL. We recommend that CD26 and CD4 be included in antibody panels to improve the detection of T-cell lymphoma/leukemias.

Most patients with ATLL have persistent MRD in the blood after treatment, indicating the need for improved therapeutics in this disease. Examination of CD4+ peripheral blood T cells for aberrant expression of CD3, CD7, and CD25 and lack of expression of CD26 by FC is highly sensitive and specific for detecting MRD in ATLL and may be applicable to other T-cell malignancies. Although in comparison with PCR for TRG rearrangement FC is slightly less sensitive, it permits quantification of residual ATLL cells and MRD and may be useful for predicting patients at risk for disease progression or relapse.


  • Supported by the Intramural Program of the National Institutes of Health, National Cancer Institute.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
View Abstract