OUP user menu

Clinicopathologic Analysis of Peripheral T-Cell Lymphoma, Follicular Variant, and Comparison With Angioimmunoblastic T-Cell Lymphoma
Bcl-6 Expression Might Affect Progression Between These Disorders

Hiroaki Miyoshi MD, Kensaku Sato MPH, Daisuke Niino MD, PhD, Fumiko Arakawa PhD, Yoshizo Kimura MD, PhD, Junichi Kiyasu MD, Masanori Takeuchi MD, Maki Yoshida MD, Yosuke Okada MD, Yukihiko Nakamura MD, Yasuo Sugita MD, PhD, Koichi Ohshima MD, PhD
DOI: http://dx.doi.org/10.1309/AJCPBPNV86VZENGV 879-889 First published online: 1 June 2012

Abstract

We examined clinicopathologic findings in 17 cases of peripheral T-cell lymphoma, follicular variant (f-PTCL), and compared these findings with angioimmunoblastic T-cell lymphoma (AITL) to determine whether they were identical to the spectrum of changes seen in AITL and how each of the findings in f-PTCL were related to the characteristics of AITL.

Almost all f-PTCL cases showed pathologic characteristics of AITL and immunohistochemical positivities in lymphoma cells for CD4, CD10, Bcl-6, PD-1, and CXCL13. Except for pathologic characteristics, clinicopathologic findings in f-PTCL had few significant differences from AITL. The positive rate for Bcl-6 expression in neoplastic cells was significantly associated with the frequency of polymorphic infiltrates, vascular proliferation, B-immunoblasts, clear cells, Epstein-Barr virus–positive lymphocytes, hepatosplenomegaly, and skin rash.

Our study confirmed the continuity between f-PTCL and AITL. Moreover, Bcl-6 expression in f-PTCL was statistically associated with the characteristics of AITL.

Key Words:
  • Peripheral T-cell lymphoma
  • Pathology
  • Follicular helper T-cell
  • Overall survival
  • Follicular T-cell lymphoma
  • Angioimmunoblastic T-cell lymphoma
  • Bcl-6

Peripheral T-cell lymphoma, not otherwise specified (PTCL, NOS) accounts for approximately 30% of peripheral T-cell lymphomas in Western countries.1 According to World Health Organization (WHO) classifications, PTCL, NOS includes 3 variants: the lymphoepithelioid, follicular, and T-zone variants.2 Lymphoma cells of PTCL, NOS, follicular variant (f-PTCL) proliferate in either a follicular lymphoma (FL)–like pattern, progressively transformed germinal center (PTGC)–like pattern, or a nodal marginal zone lymphoma (MZL)–like pattern.24

The tumor cells of f-PTCL are considered to arise from follicular helper T-cells (TFH).39 Although the pathogenesis and clinical course of f-PTCL are not well known, t(5;9) (q33;q22) translocations have been detected in some f-PTCL cases.10 In contrast, angioimmunoblastic T-cell lymphoma (AITL) is a systemic disease that clinically shows B symptoms (70%), generalized lymphadenopathy (up to 90%), Ann Arbor stage III or IV (>75%), hepatosplenomegaly (50%–80%), skin rash (50%), hypergammaglobulinemia (50%), and a median survival of less than 3 years.1113 AITL is pathologically characterized by marked proliferation of high endothelial venules, expanded follicular dendritic cell (FDC) meshworks around those venules, diffuse polymorphic infiltrates, Epstein-Barr virus (EBV)–positive B-cells, and expansion of B-immunoblasts.2 Cytogenetic and comparative genomic hybridization analyses have shown gains of 3, 5, 11q13, 18, 19, 22q, and X, and losses of 13q as typical abnormalities in AITL.11,14 The tumor cells of AITL show a phenotype of TFH as does f-PTCL.1521

Since Rüdiger et al4 first reported f-PTCL, more studies have described some similarities between f-PTCL and AITL other than their origin. Most f-PTCL cases had at least 1 pathologic feature of AITL.39 Some cases that included findings of consecutive biopsies were differently diagnosed as either f-PTCL or AITL with each biopsy.9,22 Clinical characteristics of AITL, including hypergammaglobulinemia and positive Coombs test result, were also observed in several f-PTCL cases.39 These results suggest that f-PTCL and AITL belong to the same disease spectrum. Nevertheless, this issue remains controversial because of insufficient investigations.

The tumor cells of f-PTCL and AITL reportedly originate from TFH with immunohistochemical expressions of CD10, Bcl-6, programmed death-1 (PD-1), and CXCL13.39,1521 According to previous reports of f-PTCL, tumor cells were CD10 positive in about half of the cases and Bcl-6 expression was varied.39 Almost all cases showed PD-1– and CXCL13+ cells. In contrast, CD10, Bcl-6, PD-1, and CXCL13 expression was immunohistochemically observed in 74.0%, 66.7%, 93.8%, and 97.9% of AITL cases, respectively.23 Some reports also found overexpression of these molecules at the transcriptional level in AITL.15,16 However, only a few studies have examined the relationships between these expressions and the progression of AITL.15,16

In this study, we examined the clinicopathologic findings in 17 cases of f-PTCL and compared these findings with those of AITL to determine whether they were identical to the spectrum of changes seen in AITL and how each of the histopathologic and immunohistochemical findings in f-PTCL were related to the characteristics of AITL.

Materials and Methods

Tissue Samples

We reviewed lymph node biopsy samples from 17 cases that had been submitted for diagnosis to the Department of Pathology, Kurume University, Kurume, Japan, between 2005 and 2010. Paraffin-embedded tissues and frozen samples were available. Clinical information was obtained by reviewing the patients’ medical charts. AITL data were derived from a previous study.24 Materials and clinical information were approved by the Research Ethics Committee of Kurume University and were in accordance with the Declaration of Helsinki.

Morphologic Criteria

The 17 cases included peripheral T-cell lymphomas composed of follicular growth patterns, except for Hodgkin lymphoma, anaplastic large cell lymphoma, adult T-cell leukemia/lymphoma, and other subtypes of PTCL.25 None of these cases fulfilled the morphologic criteria for AITL. The criteria for the diagnosis of AITL were based on WHO recommendations, including neoplastic clear cells, polymorphous infiltrates with eosinophils and plasma cells, an expanded FDC meshwork surrounding prominent proliferation of high endothelial venules, EBV-infected B-cells, and scattered B-immunoblasts. Two experienced hematopathologists (K.O. and D.N.) reviewed all cases.

Immunohistochemical Staining

Paraffin sections of each sample were immunostained. The antibodies [clones] used were CD4 [4B12] (MBL, Nagoya, Japan); CD20 [L-26] (Dakocytomation, Glostrup, Denmark); CD21 [1F8] (Dakocytomation); FDC [CNA.42] (Dakocytomation); CD10 [56C6] (Novocastra, Newcastle, England); Bcl-6 [P1F6] (Novocastra); PD-1 [NAT] (Abcam, Cambridge, MA); CXCL13 [goat polyclonal] (R&D, Minneapolis, MN). FDC meshworks were assessed using antibodies against CD21 or FDC. The percentage of neoplastic cells was scored as follows: –, negative (no evidence of positive cells); –/+, partially positive (5%–20% positive neoplastic cells); +, positive (>20%–40% positive neoplastic cells); ++, highly positive (>40% positive neoplastic cells).

In Situ Hybridization for EBV-Encoded RNA

EBV was detected by means of in situ hybridization with a fluorescein-conjugated EBV peptide nucleic acid (PNA) probe kit (Dakocytomation) following the manufacturer’s instructions. This probe was complementary to the 2 nuclear EBER RNAs encoded by the EBV.

T-Cell Clonality Studies

T-cell receptor (TCR) TCRCβ1 and TCRγ genes were analyzed to detect T-cell clonality. In some cases, DNA was derived from frozen samples. Rearrangements of the TCRCβ1 gene were examined using an accepted Southern blot method. For other cases, DNA was extracted from paraffin sections. Polymerase chain reaction (PCR) was used to analyze TCRγ gene rearrangements. The method was previously shown.26

Cytogenetic Analysis

The G-banding method was used for cytogenetic analysis. Karyotypes were described according to the International System for Human Cytogenetics Nomenclature (1995).

Statistical Analysis

A χ2 test was used to compare clinicopathologic features of f-PTCL and AITL24 and assess associations between clinicopathologic features in f-PTCL cases. Survival curves were calculated with the Kaplan-Meier method. A log-rank test was used for survival analysis. In statistical analysis of f-PTCL, the clinical and pathologic findings were categorized into 2 groups each, including with or without B symptoms, extranodal involvement, hepatosplenomegaly, skin rash, elevated lactate dehydrogenase (LDH) levels, hypergammaglobulinemia, therapeutic effect, mortality, growth pattern, and diffuse proliferation. Ann Arbor stage and International Prognostic Index (IPI) were classified into 4 grades. Tumor cell size was grouped into 3 grades. FDC meshworks were categorized into 3 grades: no neoplastic FDC, FDC limited in neoplastic follicles, and slightly expanded meshwork. The scoring systems of histopathologic features were classified into 4 grades as follows: (1) polymorphic infiltrates and vascular proliferation: –, not encountered; +/–, rarely encountered; +, occasionally encountered; ++, frequently encountered; (2) B-immunoblasts: –, 0 per ×200 magnification; +/–, less than 0.5 per ×200 magnification; +, 0.5 to 1.0 per ×200 magnification; ++, more than 1.0 per ×200 magnification; (3) clear cell: –, less than 5%; +/–, 5% to 20%; +, more than 20% to 40%; ++, more than 40%; (4) EBV: –, less than 0.1 per high-power field (HPF); +/–, 0.1 to 1.0 per HPF; +, more than 1.0 to 10 per HPF; ++, more than 10 per HPF.

Immunohistochemical findings were also categorized into 4 grades as described in the “Immunohistochemical Staining” section. P < .05 indicated statistical significance.

Results

Clinical Features

The 17 patients studied included 7 men (41.2%) and 10 women (58.8%) who ranged in age from 55 to 80 years (median age, 67 years). Their clinical manifestations at the time of initial biopsy are shown in Table 1.

Seven patients (41.2%) had B symptoms. Extranodal involvement, including peripheral blood and bone marrow, was observed in 8 patients (47.1%). Hepatosplenomegaly and skin rash were present in 4 (23.5%) and 5 (29.4%) patients, respectively. Thirteen patients (76.5%) had Ann Arbor stage III or IV disease. IPI was recorded as follows: 1 patient (5.9%) was classified as low risk; 3 (17.6%) were intermediate-low risk; 9 (52.9%) were intermediate-high risk; and 4 (23.5%) were high risk.

Fifteen patients (88.2%) had elevated LDH levels. Hypergammaglobulinemia was present in 8 (57.1%) of 14 patients. Fourteen patients (82.4%) had received chemotherapy, including a cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) regimen; a cyclophosphamide, pirarubicin, vincristine, and prednisone (THP-COP) regimen; or a vincristine, prednisone, etoposide, and chlorambucil (OPEC) regimen. Steroids only were administered to 1 patient (5.9%). Two patients (11.8%) were followed up without treatment.

Of the 14 patients who had received chemotherapy, 10 (71.4%) achieved a complete response (CR) or a complete response uncertain (CRu), while 4 patients (28.6%) had a partial response (PR). Relapse after CR was observed in 3 patients.

Four patients (23.5%) died of their disease or disease-related complications; 10 patients (58.8%) were still alive at the time of this writing, including 6 with sustained CR or CRu; and 3 patients (17.6%) dropped out after a median follow-up period of 13 months (range 1 to 96 months).

View this table:
Table 1

Histopathologic Findings and Immunohistochemical Analysis

Table 2 shows the initial biopsy findings for the 17 cases and consecutive biopsy findings for case 5 performed at the time of relapse. Some pathologic findings are shown in Image 1. Seven initial biopsy specimens (41.2%) had no FDC meshwork associated with neoplasm (listed as Absent or Residual). An FDC meshwork limited within neoplastic follicles was observed in 6 (35.3%) cases and a slightly expanded meshwork was found in 4 cases (23.5%). Seven cases (41.2%) had polymorphic infiltrates, including plasma cells and eosinophils, while the infiltrates were observed only within intravascular lesions in 1 case. Thirteen patients (76.5%) had mild to moderate vascular proliferation.

CD20+ large B-immunoblasts were seen in 14 cases (82.4%), up to 5.9 counts per ×200 magnification. Fifteen cases (88.2%) had clear cells in various proportions (average, 28.8%; range, 0–70%). EBV-infected lymphocytes were observed in 14 cases (82.4%). Lymphoma cells were small in 2 cases (11.8%), medium-sized in 8 cases (47.1%), and large in 7 cases (41.2%). Eight cases (47.1%) showed sharply defined follicles formed by neoplastic cells mimicking follicular lymphoma (FL-like).

Nine cases (52.9%) had large and regular or irregular nodules reminiscent of progressively transformed germinal centers (PTGC-like). Partial diffuse proliferation was observed in 4 cases (23.5%). Lymphoma cells were positive for CD4 in all cases, whereas cells were CD10 positive in 6 cases (35.3%). More than half of all cases were positive or highly positive for Bcl-6 (58.5%; 10/17), PD-1 (68.8%; 11/16), and CXCL13 (76.5%; 13/17).

The consecutive biopsy specimen from patient 5 had larger lymphoma cells and a PTGC-like growth pattern, and appeared to be closer to the pathologic features of AITL than the initial biopsy specimen.

Genotypic Analysis

TCR gene rearrangement analyses were performed in all cases except for case 8. Four cases showed clonal TCRCβ1 gene rearrangements with the Southern blot method. PCR was used to analyze TCRγ gene rearrangements in the other 12 cases. Clonality was observed in 10 cases, but was not detected in 2 others (cases 6 and 11).

Cytogenetic Analysis

Cytogenetic analysis using the G-banding method was performed for 16 cases Table 3. In 3 cases, the results were not informative because analyzable metaphase chromosomes could not be obtained in 2 patients (cases 1 and 14) and more than 1 abnormality was not observed in 1 patient (case 10). Eight cases showed normal karyotypes. However, this may not have reflected the karyotypes of lymphoma cells, but rather those of reactive cells. In 5 cases the same abnormal karyotypes was found in more than 2 cells. Gains of 3, 5, 18, and/or 19 were observed in 4 patients (cases 2, 3, 5, and 16; case 5 had a relapse). The t(5;9)(q33;q22) translocation was not observed in any of the cases.

Comparison of Clinicopathologic Features With AITL

Table 4 compares clinicopathologic findings of f-PTCL and AITL. In clinical features, the cases of f-PTCL had significantly lower frequency in B symptoms (P = .048) and skin rash (P = .014). The pathologic features characteristic of AITL tended to be observed less frequently in the cases of f-PTCL except for EBV-positive lymphocytes. The expression of TFH markers also indicated the same results despite no significant differences.

Figure 1 shows the overall survival curves for the current cases compared with AITL cases. The patients with f-PTCL tended to have a better prognosis than those with AITL, but the difference was not significant (P = .0521).

Statistical Analysis of Clinicopathologic Features

Possible associations between clinicopathologic findings from initial biopsies were statistically analyzed. The main results are shown in Table 5. Cases with a higher proportion of Bcl-6– and/or CD10+ neoplastic cells had significantly more clinicopathologic findings characteristic of AITL. The cases with PTGC-like patterns had more features of AITL than those with FL-like patterns. An FDC meshwork tended to show similar outcomes.

Associations other than the results in Table 5 were as follows. Associations between CD10 and Bcl-6 expressions (P = .0057) and between Bcl-6 and PD-1 expressions (P = .0369) were significant. The positive rates for PD-1 and CXCL13 were associated with vascular proliferation (P = .034 and .0168, respectively). The cases with PTGC-like patterns tended to have higher IPI scores than those with FL-like patterns. EBV-infected lymphocytes were associated with polymorphic infiltrates (P = .0416) and mortality (P = .0278). A significant association was also observed between polymorphic infiltrates and mortality (P = .0083). Cases with diffuse proliferation, more clear cells, and larger lymphoma cells had higher frequencies of skin rash (P = .0263, .017, and .0161, respectively) and hypergammaglobulinemia (P = .0263, .017, and .0161, respectively).

Discussion

This study showed no significant differences between f-PTCL and AITL in clinicopathologic features except for the pathologic characteristics of AITL. This result confirms the results of previous studies that f-PTCL possibly represents a peculiar stage of AITL. Bcl-6 expression in f-PTCL is statistically associated with the clinicopathologic characteristics of AITL, but the expression was not significantly different between f-PTCL and AITL.

In previous studies of f-PTCL,39 most cases had at least 1 pathologic feature of AITL, while a diffuse effacement of the architecture was infrequent. Varying percentages of f-PTCL cases also presented clinical features of AITL. Our study showed results similar to those of other studies and emphasized the relationships between f-PTCL and AITL. Similar to previous studies, we found partial features of AITL in most cases, pathologic progression on multiple biopsies in 1 case, pathologic transition from a follicular growth pattern to diffuse proliferation in 4 cases, characteristic chromosome abnormalities of AITL in 4 cases, and few significant differences between clinicopathologic features of f-PTCL and AITL including survival curves. Although the survival curves were not significantly different, the P value was .0521. A longer follow-up period might show a lower P value and result in a significant difference. More studies with longer follow-up periods should resolve this problem.

Image 1

Morphologic and immunohistochemical findings in peripheral T-cell lymphoma, follicular variant (f-PTCL). A (Case10), Numerous pale-pink follicles (FL-like) composed of lymphoma cells (×20). B (Case 17), Large vague follicles (progressively transformed germinal center [PTGC]–like) (×12.5). C (Case 6), Irregular follicles of various sizes (PTGC-like) and diffuse proliferation (at the right edge of the image) (×12.5). D (Case 10), Scattered residual follicular dendritic cells (FDCs) (arrows) (×20). E (Case 17), Limited FDC meshworks in neoplastic follicles (×20). F (Case 6), FDC meshworks in neoplastic follicles and mildly expanded FDC meshworks (at the right edge of the image) (×20). G (Case 4), Small lymphoma cells that were positive for CD4 (×400). H (Case 17), Medium-sized lymphoma cells, more than half of which were positive for Bcl-6 (×400). I (Case 6), Large neoplastic cells with clear cytoplasm, most of which were positive for PD-1 (×400). J (Case16), Lymphoma cells positive for CXCL13 and polymorphous infiltrates, including eosinophils and plasma cells (inset, ×200; ×400). K (Case 4), Vascular proliferation (×100). L (Case 2), CD20-positive immunoblasts (×400). (A–C, G (insert), H, I (inset), J, H&E; D, F, immunohistochemical staining of FDC; E, CD21; G, CD4; H (inset) Bcl-6; I, PD-1; J, CXCL13; K, periodic acid–schiff hematoxylin; L, CD20.)

The association between tumor cell size and progression is less well established in T-cell lymphomas than in B-cell lymphomas. In general, the tumor cells of AITL are medium sized.2 Some reports have shown large tumor cells in AITL and transformations from medium-sized cells to large cells were observed in several cases.27,28 Although most f-PTCL cases involved medium-sized cells,37 a few cases had large tumor cells.8 In this study, large tumor cells were observed in about half of the f-PTCL cases and tended to be higher in proportion in the f-PTCL cases than in the AITL cases. Tumor cell size was significantly associated with the expressions of CD10 and Bcl-6, skin rash, and hypergammaglobulinemia, but most other findings were not associated with tumor cell size. It is unlikely that tumor cell size clearly reflects the disease progression from f-PTCL to AITL.

View this table:
Table 3
View this table:
Table 4

AITL reportedly originates from germinal center TFH cells with the immunohistologic expressions of CD4, CD10, Bcl-6, PD-1, and CXCL13.1720 CD10 is a 100-kDa, type II, integral membrane protein with neutral endopeptidase activity.29 The expression of CD10 is postulated to regulate apoptosis by interfering with negative and positive signals from the extracellular environment.30 Bcl-6 is considered to be a transcriptional factor that is selectively expressed in TFH cells and required for their differentiation.31,32 PD-1 is a member of the CD28 family of receptors that includes CD28, cytotoxic T lymphocyte–associated antigen 4, and B- and T-lymphocyte attenuator. These receptors have roles in regulating cellular immune responses.33 CXCL13 is a chemokine specifically expressed in lymphoid follicles and required for B cells’ entry into germinal centers.34,35 The overexpression of these molecules at the transcriptional level was also found in AITL with a TFH origin.15,16

Figure 1

Overall survival curves for peripheral T-cell lymphoma, follicular variant (f-PTCL) and angioimmunoblastic T-cell lymphoma (AITL). Although patients with f-PTCL tended to have a better prognosis than those with AITL, the difference was not significant (P = .0521).

f-PTCL has been similarly described to arise from TFH cells. Previous studies have shown that almost all cases of f-PTCL were positive for CD4. CD10 was expressed in about half of the cases. The Bcl-6 positive rate ranged from less than half to three-quarters of f-PTCL cases. PD-1 and CXCL13 were observed in almost all cases.39 The results of our study were consistent with these studies.

In the present study, the expression of Bcl-6 in f-PTCL was most frequently associated with the clinicopathologic features of AITL despite no significant difference in comparison between f-PTCL and AITL. This result suggests the possibility that Bcl-6 may play an important role in the transition between f-PTCL and AITL. In B-cell lineages, the function and pathogenesis involving Bcl-6 have been investigated. Bcl-6 promotes cell proliferation by repressing a cell cycle inhibitor.36 The terminal differentiation of germinal center B cells to plasma cells requires the expression of PRDM1 that is repressed by Bcl-6.37 DNA repair and p53 activity are inhibited by Bcl-6.38 Bcl-6 enhances B-cell receptor signaling in normal B cells.39 Some of these activities can cause malignant alterations if they are excessive.

The overexpression of Bcl-6 has been shown in some cases of diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma, and nodular lymphocyte predominant Hodgkin lymphoma through chromosomal translocations or disruption of regulatory regions within the Bcl-6 gene.40 Compared with B-cell lineages, far fewer studies have analyzed the function of Bcl-6 in T-cell lineages, including TFH. Bcl-6 expression by TFH cells is upregulated by interleukin (IL)-6 and IL-21. Overexpression of Bcl-6 induced TFH-related gene expression, including CXCR5, a receptor for CXCL13, and inhibited the differentiation of other helper T-cell lineage cells.31,32 It is not well known if and how the overexpression of Bcl-6 can affect AITL pathogenesis. However, the activities of Bcl-6 for promoting cell proliferation and suppressing DNA repair may be associated with the progression of AITL; for example, a higher expression of Bcl-6 can lead to a more aggressive clinical course in breast cancer.41 Further studies are needed to confirm the relationship between Bcl-6 expression and the transition from f-PTCL to AITL.

Streubel et al10 first showed translocations on chromosomes 5q33 and 9q22 in some f-PTCL cases that resulted in an ITK-SYK fusion transcript. ITK belongs to a group of 5 related kinases comprising the Tec family and affects T-cell development and T-cell receptor signaling.42 SYK is a nonreceptor protein kinase that is a key regulator of multiple biochemical signal transduction pathways.43 Huang et al9 also found t(5;9)(q33;q22) translocations in 4 of 30 f-PTCL cases using fluorescence in situ hybridization (FISH). Although no such translocations were found in other studies,4,5,7 possible translocations were assessed using cytogenetic analysis and not FISH. In the present study, cytogenetic analysis showed characteristic abnormalities of AITL in f-PTCL cases but no t(5;9)(q33;q22) translocations. On the other hand, this translocation was not observed in AITL.10,44 The significance of this translocation and its specificity remains controversial. More investigations that include genetic studies of f-PTCL and AITL are required.

View this table:
Table 5

In conclusion, our clinicopathologic analysis of f-PTCL cases and comparison between f-PTCL and AITL cases showed continuity between f-PTCL and AITL. Bcl-6 expression in f-PTCL cases was statistically associated with the characteristics of AITL, but the expression was not significantly different between these disorders. These results suggest that Bcl-6 expression possibly affects the transition between f-PTCL and AITL. More investigations, including genetic and transcriptional studies, are required to identify the role of Bcl-6 in the pathogenesis of f-PTCL and AITL.

References

  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.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
View Abstract