OUP user menu

Molecular Pathology of Breast Cancer
What a Pathologist Needs to Know

Kimberly H. Allison MD
DOI: http://dx.doi.org/10.1309/AJCPIV9IQ1MRQMOO 770-780 First published online: 1 December 2012

Abstract

Pathologists are now more than ever “diagnostic oncologists” and serve a critical role as clinical consultants on the biology of disease. In the last decade and a half, molecular information has transformed our thinking about the biologic diversity of breast cancers and redirected the way clinical treatment decisions are made. A basic understanding of the current molecular classification of breast cancers and the biologic pathways from precursors to invasive disease is key to both informing diagnostic practice and serving as clinical consultants. In addition, both single-marker and panel-based molecular tests are currently being utilized in breast cancer tissue to predict the benefit of specific therapies such as HER2-targeted biologic therapy and chemotherapy. Familiarity with the current issues involving these molecular tests as well as the pathologist’s role in ensuring appropriate tissue handling, tissue selection, and results interpretation and correlation are paramount to providing optimal patient care.

Key Words:
  • Molecular diagnostics
  • Breast
  • HER2 FISH testing
  • Oncotype DX

When it comes to breast cancer diagnostics, pathologists have become the translators of many layers of biologic information including morphologic features, protein and gene expression, and genetic information. Molecular techniques have both changed our understanding of the basic biology of breast cancer and provided the foundation for new methods of “personalized” prognostic and predictive testing. While traditional staging on the basis of tumor size and lymph node status remains the cornerstone of outcome indicators, it has become clear that not all breast cancers presenting at the same stage have the same underlying biology or clinical behavior.1,2 While morphologic assessment of factors such as Nottingham grade and histologic subtype offer prognostic information about how aggressively a cancer may behave clinically, the biology of each cancer can additionally be resolved by combining morphologic factors with results of ancillary tissue-based testing.

As pathologists and as “diagnostic oncologists,” we are the experts in tissue-based testing and will be looked to for both clinical guidance as well as leadership in test quality and innovation. Therefore, it is critical to have a current understanding of both the benefits and limitations of molecular testing in the clinical setting. This review will cover (1) how molecular testing has changed our current understanding of the biology of breast cancer and how this knowledge can be used to inform clinical diagnosis; (2) how molecular testing is currently used to inform clinical practice and treatment decisions in breast cancer; and (3) how pathologists can serve as molecular testing consultants and help guide the clinical decisions related to them.

Molecular Testing as a Tool for Understanding the Biology of Breast Cancer

The explosion of molecular information in the past decade and a half has led to a better understanding of the biologic diversity of breast cancers as well as clues to the different etiologic pathways to breast cancer development. As pathologists, we are expected to serve our clinical teams as consultants on the biologic understanding of breast cancer subtypes and their pathogenesis.

Molecular/Intrinsic Subtypes

It is useful to establish groupings or subtypes of breast cancers that have both biologic and clinical relevance. With more than 80% of breast cancers classified histologically as “invasive ductal carcinoma, not otherwise specified,” there is clearly more biologic diversity among breast cancers than our histologic classification scheme alone implies. Using additional factors such as Nottingham grade, hormone receptor status, and HER2 status, breast cancers can be categorized by both biology and therapeutic opportunities.

Gene expression profiling has reconfirmed that the major drivers of breast cancer biology are hormone receptor–related genes, HER2-related genes, and have highlighted the importance of proliferation-related genes.38 Seminal studies evaluating the gene expression profiles of breast cancers have segregated them into “intrinsic subtypes” or “molecular subtypes” based on the relatedness of their gene expression patterns using a hierarchical clustering.911 These data support 4 main molecular/intrinsic breast cancer subtypes that have prognostic relevance to survival. These subtypes have been termed luminal A, luminal B, HER2-related, and basal-like (a fifth “normal breast–like” category has not been reproducibly defined). Figure 1 summarizes the characteristics of these molecularly defined main subtypes. These 4 subtypes, originally defined by gene-expressing profiling, were also recently demonstrated as valid groupings using different platforms, including genomic DNA copy number arrays, DNA methylation, exome sequencing, microRNA sequencing, and reverse-phase protein assays.12 At the most basic level, the luminal subtypes share expression of estrogen receptor (ER)–related genes and have better overall survival than the HER2-related and the basal-like subtypes, which are typically (but not uniformly) ER negative.10,13

Studies correlating the molecular subtypes of breast cancer with the more traditional, slide-based pathologic features of the same cancers have identified correlations between the 2.14,15 Luminal cancers express hormone receptors and are lower grade while the HER2 subtypes overexpress HER2 gene products and are higher grade. Luminal B cancers have a worse prognosis than luminal A cancers; often have lower expression levels of hormone receptors, higher Nottingham grade, and higher proliferative rates; and can be HER2 positive.16 There is clinical interest in distinguishing the luminal B cancers from luminal A cancers because they may be a subset of ER-positive cancers that derive benefit from more aggressive therapy.17 However, it is also acknowledged that the differences between these 2 groups are largely based on differences in proliferation-related genes and, rather than representing distinct subtypes of breast cancer, are more likely 2 ends of a spectrum of ER-positive disease.3,18,19

The basal-like subtype appears to overlap substantially with the population of breast cancers that are “triple negative” (ER/progesterone receptor (PR) and HER2 negative) and high grade but also are associated with characteristic histologic features such as solid-pushing borders, geographic areas of necrosis, and dense lymphocytic infiltrates.2025 Cancers with a basal-like molecular profile can be identified with high specificity (but only about 75% sensitivity) by the following immunohistochemical profile: ER and PR negative, HER2 negative, and cytokeratin (CK) 5/6 or epidermal growth factor receptor (EGFR) positive (although other variations of this panel have also been used).14,22,24,26,27 However, clinically validated thresholds for CK5/6 or EGFR staining are still lacking. In addition, there is currently no difference in the standard treatment of a triple-negative and a basal-like breast cancer; therefore definitive classification of a basal-like cancer has not become standard clinical practice for drug therapy–related decisions.28,29

Clinically, basal-like cancers are noted to occur more often in younger patients and African American women and are associated with a worse prognosis.30,31 Interestingly, the vast majority of BRCA1-associated breast cancers appear to have a basal-like molecular profile, suggesting a common pathway of carcinogenesis in these patients.3235 However, a basal-like or triple-negative profile by itself does not necessarily predict BRCA1 mutation status (ie, many of these cancers are in BRCA1-negative patients).36,37

The clinical usefulness of molecular subtype testing of individual breast cancers is still unclear, but assays to classify clinical cases into subtypes have recently become commercially available. The PAM-50 Breast Cancer Intrinsic Classifier assay (NanoString, ARUP Laboratories, Salt Lake City, UT) is a reverse transcription polymerase chain reaction (RT-PCR) assay that uses the gene product levels of 50 genes thought to be inherent to the molecular subtype.38,39 It can be performed on formalin-fixed, paraffin-embedded tissue and categorizes individual cancers into the molecular subtype to which they are most similar. The MammaPrint test (Agendia, Irvine, CA), which currently requires fresh tissue specimens because of its microarray-based platform, also includes an option to test for molecular subtype (BluePrint) in its panel of tests.40 However, these assays have been criticized as single-sample predictors of molecular subtypes because they were developed on the basis of hierarchical clustering rather than as a predictive test on individual samples.41 In fact, the results of molecular classification testing on a case-by-case basis is highly dependent on the platform and data evaluation techniques used, with only moderate agreement among them.4246 While the classification of the basal-like subtype appears to be the most reliable, the additional subtypes can be variable.25,26 The impact of intermixed normal tissue has also been suggested as a source of interference with gene expression profiling used as a predictive single-sample test.25,36,37,47,48

Figure 1

Summary of the features of the basic molecular/intrinsic breast cancer subtypes. The characteristics of these categories have been generalized for simplification. * Notable exceptions to the typically high-grade histologic features associated with the basal molecular subtype are lower-grade cancers of the following special types: adenoid cystic carcinomas, low-grade metaplastic carcinomas, apocrine carcinomas, and invasive carcinomas associated with microglandular adenosis. CK, cytokeratin; EGFR, epidermal growth factor; ER, estrogen receptor; PR, progesterone receptor.

Although molecular/intrinsic subtypes have emphasized the importance of the biology driving different breast cancers, it remains to be seen whether molecular assays for subtyping will prove to be reproducible, clinically useful, and practical. However, as discussed in the following sections, knowledge of these categories and how they are defined can inform diagnosis in clinical practice.

Molecular Etiology of Breast Cancer

Molecular evidence has shed light on the different pathways leading to the development of invasive breast cancers. Studies looking at the patterns of gene copy number changes and mutations present have identified certain genetic alterations (such as deletions of 16q and gains of 1q) in ER-positive cancers that appear to be rare in ER-negative cancers.4954 ER-negative cancers tend to harbor more severe genetic aberrations such as p53 mutations, HER2 amplifications, BRCA1 dysfunction, and high genomic instability. This evidence suggests that ER-positive and ER-negative cancers have 2 distinct pathways of development that do not commonly overlap.

In addition, contrary to initial data suggesting that progression from low- to high-grade invasive cancers was rare, it appears that the ER-positive subset of cancers may progress from low to high grade.50,55 Approximately 50% of grade 3 ER-positive cancers have the same 16q and 1q alterations as grade 1 ER-positive cancers, suggesting a common pathway to development.50 Grade 3 ER-positive cancers tend to have accumulated additional genetic changes resulting in higher proliferation rates and additional genomic instability. Figure 2 shows an overview of these proposed pathways in breast cancer development.

Recent evidence also suggests that in situ and invasive ER-positive ductal and lobular cancers are nearly identical from a genetic standpoint, with the same 16q and 1q alterations but the additional loss of E-cadherin expression in the lobular phenotype. In fact, the molecular similarities among flat epithelial atypia, atypical ductal hyperplasia, and lobular in situ neoplasias all point to a very similar low-grade pathway of neoplastic progression and explain their frequent association with each other on histology.51,5662 In contrast, basal-like breast cancers are frequently noted to lack a recognizable in situ component. Although initially thought to arise from the basal/myoepithelial cell of the breast, recent evidence refutes this theory but sheds little light on other possible precursors.63 A high-grade, triple-negative form of ductal carcinoma in situ (DCIS) has been recognized, and microglandular adenosis has also been linked as a possible precursor, but both are much less prevalent than invasive triple-negative/basal-like breast cancers.6470 Extensive high-grade DCIS (often of the so-called “comedo type”) is more frequently associated with HER2-positive invasive disease.

Lastly, recent evidence highlighting the molecular and genetic heterogeneity of cancers (including breast) has brought to the forefront the concept of clonal evolution in cancer.7178 This concept recognizes that different subpopulations that have acquired additional genetic mutations compete with each other under selective pressure as cancers grow, progress, and metastasize.79 An extensive review of this topic is beyond the scope of this article, but the concept of clonal evolution can be incorporated into the current classification of molecular subtypes in breast cancer by recognizing that some cancer subtypes are more rapidly evolving because of their inherently high genetic instability (ie, the basal-like subtype). It follows that with disease progression, any cancer’s clinically relevant phenotype may change. Changes in the hormone receptor or HER2 status with metastatic progression are estimated to change approximately 10% to 40% of the time.80,81 These findings may have various alternative explanations, including the way in which non–breast tissue is handled. However, because of the effect on therapeutic options for a patient with metastatic disease (especially in the setting of disease previously negative for these markers), the current National Comprehensive Cancer Network (NCCN) recommendations are to retest metastatic disease for hormone receptors and HER2 status.82 The implications for reporting HER2 heterogeneity by means of fluorescence in situ hybridization (FISH) testing are also relevant to this topic and will be discussed.

The aforementioned combination of concepts recognizes the common molecular pathways in breast cancer development and progression as well as the inherent plasticity of cancer that makes biology reasonably predictable when looking at large groups but less predictable on an individual case basis.

Figure 2

A model of the molecular pathogenesis of breast cancers. The more common pathway is shown at the top, with increasingly less common pathways shown below. ER, estrogen receptor.

Correlating Molecular Information With Morphology to Inform Diagnostic Practice

Recognition of the histologic features that correlate with molecular subtypes and what we know about etiologic pathways can assist in clinical diagnostic categorization. Characteristics classically associated with these subtypes should serve as a check and, if they are discordant, should prompt reevaluation of the details of the diagnosis, the validity of prognostic/predictive markers on the case, as well as consideration of recognized exceptions to the classic features of the intrinsic subtypes. Well-characterized exceptions to the descriptions of subtypes listed in Figure 1 include low-grade triple-negative cancers with unique histologic subtypes such as adenoid cystic carcinoma, apocrine carcinoma, microglandular adenosis–associated cancers, and lower-grade metaplastic carcinoma variants (fibromatosis-like, squamous, and adenosquamous variants), all of which have basal-like molecular profiles but, unlike most cancers with this profile, are classically low grade.8386 Using this approach, recognition of discordant features (such as low grade but ER negative) can aid in reconsideration of histologic type, grade, and the results of ancillary tests.

Knowledge of the relatedness on the molecular underpinnings of precursor lesions can also aid in recognizing a constellation of findings that make biologic sense in a given case. For example, the presence of low-grade precursor lesions such as flat epithelial atypia, atypical ductal hyperplasia, or atypical lobular hyperplasia/lobular carcinoma in situ may prompt a more careful examination for low-grade invasive lesions with subtle histologic findings such as small invasive lobular or tubular carcinomas. In contrast, a case with extensive high-grade, ER-negative DCIS may harbor small foci of invasive HER2-positive disease (which can often be highlighted by HER2 or pancytokeratin immunohistochemistry [IHC]).

The Future of Molecularly Defined Subtypes of Breast Cancer: A Continuing Evolution

The major gene expression subtypes serve as useful categories, but given the current therapeutic options, when genomic data are incorporated into the picture, these categories can become even more complex; this suggests that the molecular classification of breast cancers is still evolving. Curtis and colleagues recently analyzed both the genomic and transcriptomic profiles of more than 2,000 fresh frozen breast cancer samples and found not 4 but 10 biologically distinct subgroups with correlations to outcome.87 In addition, data looking at the genomic alterations in breast cancers indicate that although approximately half of all driver mutations are found in more than 10% of breast cancers, the other half are relatively unique to each cancer (present in <10%).88 This complexity has major implications for the development of tests to identify candidates for therapy and may result in blurring the lines between cancer types as we identify common targets in cancers arising from different organs.

Molecular Testing in Clinical Practice and Treatment Decisions

Single-Marker Molecular Testing for Prediction of Response to Therapy: HER2

Traditionally, ancillary tissue-based testing has used a single marker at a time. It is currently standard practice to test breast cancer tissue for ER and PR using IHC.89 These hormone receptor IHC markers correlate with benefit from endocrine therapy and are indicators of overall prognosis.

HER2 testing is also standard practice because it predicts response to HER2-targeted therapies as well as a likelihood of response to certain chemotherapeutic regimens and prognoses. HER2 testing is performed most commonly with either IHC or molecular methods such as FISH. Additional in situ molecular methods have been approved and used, such as chromogenic in situ hybridization and silver staining in situ hybridization. Because of the clinical impact of this test and reports of high discordance rates among laboratories, the details of HER2 testing and reporting (and now also ER/PR testing) became the focus of specific guidelines put forth by the College of American Pathologists (CAP) and the American Society of Clinical Oncology (ASCO) in 2007.90 These guidelines make recommendations on appropriate tissue handling, test methodology, validation, scoring, and reporting, and adherence to certain aspects are CAP required. This regulation is likely to continue to expand to any laboratory test that is used to offer a patient a specific therapeutic option (whether molecular or not).

Despite guidelines, there are still grey zones in HER2 testing. HER2 FISH results in the equivocal range (mean HER2/CEP17 ratio of 1.8–2.2 or 4–6 absolute HER2 signals/cell), which occur in an estimated 5% of cases, still have unclear treatment implications, but clinicians will often seek pathologist consultation in this setting.91 CAP has issued a clarification statement that all cases with a ratio of 2.0 or more were considered eligible for HER2-targeted therapy in several of the initial trials conducted and that this threshold should still be used to guide clinical management.92 But repeat testing by a second method (such as IHC) or on an additional tissue specimen is also recommended in the CAP/ASCO guidelines for cases in the equivocal range due to proximity to the 2.0 threshold.

In addition, in recognition of possible genetic heterogeneity for HER2 gene amplification on FISH testing, a CAP expert panel issued additional recommendations for reporting HER2 heterogeneity found on FISH.90 However, the frequency of heterogeneity for HER2 gene amplification by their proposed criteria (5%–50% of individual cells amplified) appears to be much higher than early limited evidence suggested (as many as 20%–30% meet the proposed criteria, most of which have only 5%–15% amplified cells), and these criteria also result in many nonamplified cases being classified as heterogeneous. Therefore this proposal has been questioned and not widely adopted.9399 Interestingly, the evidence suggests that FISH equivocal cases often display significant percentages of amplified cells (30%–50%), and therefore this may be an appropriate setting in which to report the presence of heterogeneity and consider treatment.93,99 However, evidence is still lacking regarding which thresholds are clinically meaningful.

Molecular Signatures for Prediction and Prognostication

In contrast to the single-marker approach, panel-based molecular testing has been used more recently to develop tests that predict both prognosis and response to therapy.100 Although hormone receptor and HER2 tests have been clinically validated to predict which patients will respond to hormone and HER2-targeted therapies, there has been great interest in tests that will predict response/need for chemotherapy as well. Various calculators or algorithms, such as ADJUVANT! Online (www.adjuvantonline.com), that determine chemotherapy benefit based on clinical and pathologic features are available.101 There is clear benefit to treating groups with multiple high-risk features (ie, younger age, lymph node positive, high grade, ER negative) with chemotherapy. However, there is less clear benefit to treating patients with traditionally low-risk features (ie, lymph node negative, ER positive). Therefore, gene expression profiles have been developed as a way to tease out differences in prognosis and chemotherapy benefit in this group. Although there are many signatures in various phases of development, Oncotype DX (Genomic Health, Redwood City, CA) and MammaPrint are currently the most common clinically used tests. Table 1 summarizes the features of these tests.

Both Oncotype DX and MammaPrint offer prognostic signatures in patients with breast cancer but with slightly different target populations and different testing platforms.102,103 Both have been well validated retrospectively using tissue specimens from clinical trials, but their clinical usefulness is still being more rigorously evaluated in prospective studies.103112 MammaPrint segregates all breast cancers into low- and high-risk profiles using a microarray-based gene expression profile focused on 70 genes.113,114 Oncotype DX was developed for use in node-negative, ER-positive cancers (although it is also used in the node-positive setting). RT-PCR levels of 16 cancer-related gene products are used to calculate a recurrence score (RS) that is reported as a continuous variable with stratification into low-, intermediate-, and high-risk categories. Because of its microarray platform, MammaPrint requires fresh tissue, which has somewhat restricted its clinical usefulness. Oncotype DX can be used on formalin-fixed paraffin-embedded tissue, allowing the test to be run on archived tissue blocks at any point in clinical decision making.115 In addition, Oncotype DX has validated its RS to both overall outcome as well as prediction of chemotherapy benefit.109 Therefore, the current NCCN guidelines advise oncologists to withhold chemotherapy in ER-positive, lymph node–negative patients with a low Oncotype DX RS and offer it if the result is in the high-risk category. But an intermediate risk result is still considered a treatment gray zone. Results from the Trial Assigning Individualized Options for Treatment (TAILORx) hope to further stratify this group.110

The cost of these assays is relatively high (approximately $3,500 for Oncotype DX and $4,000 for MammaPrint) compared with traditional pathology testing, and the benefit over traditional clinicopathologic predictive factors may be limited to cancers with intermediate features. The RS calculation used in Oncoytpe DX testing is heavily weighted for proliferation-related markers but also includes markers used routinely in diagnoses such as ER, PR, and HER2. Although evidence shows that tests such as Oncoytpe DX more accurately predict outcome and spare more women from chemotherapy than algorithms such as ADJUVANT! Online, other evolving evidence suggests that different algorithms may be used to predict RS and outcomes that take into account more of the pathology-based factors such as levels of ER, PR, and HER2, grade, and proliferation index markers such as Ki67.106,116122

View this table:
Table 1

Combining molecular prognostic and predictive signatures with traditional studies can add further data to the treatment-related decision-making process. However, it is important to remember that molecular testing is not necessarily a new “gold standard.” One benefit of more traditional in situ tests such as IHC is the visual confirmation that scoring is only performed on the invasive cancer. With molecular-based multiplexed tests, intermixed inflammatory cells, carcinoma in situ, and normal tissue may influence results.47,48,118 In addition, many of these proprietary tests are only performed in a single clinical laboratory without the benefit of external confirmation of results on a routine basis. Pathologists are responsible for selecting material to be sent to these commercial laboratories for molecular testing and should be cognizant of unusual results as well as which material is most appropriate to send (recommendations for which are listed herein).

How Pathologists Can Help Ensure Molecular Testing Accuracy

Molecular and nonmolecular tissue-based testing can be effected via many variables. Pathologists, as experts in both tissue morphology and ancillary testing, must strive to ensure the accuracy of molecular-based tests performed on tissue samples used either in their own laboratory or sent to another laboratory. This may occur by ensuring appropriate tissue handling before an assay is performed or blocks are selected for send-out tests, or may involve review and correlation of results with clinicians. These recommendations are summarized in Figure 3.

Control of Preanalytic Variables: Ischemic Time and Fixation

Most molecular tests are susceptible to preanalytic variables such as ischemic time because of potential degradation of liable targets.123 The CAP/ASCO HER2 and ER/PR testing guidelines currently recommend that ischemic time be as minimal as possible and preferably less than 1 hour.89,90 Ensuring that this happens requires coordination between the clinical team acquiring the specimen (typically radiology or surgery) and pathology and is a conversation that pathology often needs to initiate. The often immediate fixation of breast needle core biopsies can make them a preferable sample for some tests compared with surgically acquired specimens that may have more prolonged ischemic time. However, limited material may be present in core biopsy samples, so standard procedures are recommended to ensure that surgically acquired samples also receive appropriate handling.

For molecular testing that can be performed on formalin-fixed tissue, the type of fixative and time in fixative are significant variables. CAP/ASCO recommends a minimum of 6 and a maximum of 48 hours in neutral-buffered formalin for HER2 FISH testing.89,90 Although these time points have not been extensively validated, they are a starting point for standardization, and adherence to these recommendations should be targeted with documentation when reporting cases that are outliers because of potential effects on molecular and nonmolecular testing.

Figure 3

A model for the pathologist’s role in molecular testing of breast cancers.

Selection of Blocks for Testing

For laboratories performing HER2 FISH testing, it is required to circle the area with the invasive carcinoma so that technologists know in which areas to count signals. To appropriately screen for amplified populations or subpopulations, as much of the invasive carcinoma should be circled as possible, with attention being paid to the exclusion of areas of carcinoma in situ. This can be done on the H&E-stained slide, or if performed with concurrent HER2 IHC the IHC-stained slide can be circled. The latter can be especially useful when there are distinctly different levels of HER2 expression on IHC in different areas of the slide. When a distinctly clustered subpopulation appears to have higher HER2 protein expression and these findings match clustered areas of HER2 gene amplification on FISH, these areas can be counted and reported as a separate HER2-amplified subpopulation.124,125 Pathologists will have to guide technologists regarding where to count each population in these cases and review findings to ensure appropriate scoring.

Requests for block selection for additional testing, such as Oncotype DX, are frequently made to pathology. The block with invasive cancer of the worst grade and phenotype should be offered unless otherwise guided by the requesting clinician. Ideal block selection would avoid substantial amounts of carcinoma in situ, biopsy site changes, and inflammation because these may influence results.

Review and Correlation of Results With Histologic Findings

Finally, the results of molecular tests should be correlated with the original histologic findings. For example, an unexpected result might be a high Oncotype DX score or a HER2-positive FISH result in a low-grade invasive cancer. Re-review of the material sent for additional testing to look for potential confounders such as inflammation, carcinoma in situ, etc, as well as a reevaluation of the initial histologic findings, can be useful in identifying causes for the unexpected result.

Conclusions

Molecular testing in breast cancer is still in evolution. With the increasing use of more sophisticated techniques such as deep sequencing, we may continue to discover more high-throughput ways of harnessing large amounts of data to predict outcomes and treatment responses in breast cancer. However, moving a test from the bench to the bedside is a long process that requires robust validation. For a new assay to become standard practice, in addition to the test validation required, it must be clinically useful, practical, and cost-effective. Because of this challenge, some tests, despite using very sophisticated molecular techniques, may remain most useful in a research setting and not prove to be clinically essential. However, there is little doubt that molecular testing is changing, and will continue to change, the way we diagnose and treat breast cancer.

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.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
View Abstract