SEARCH

SEARCH BY CITATION

Keywords:

  • Lynch syndrome;
  • endometrial cancer;
  • endometrial hyperplasia;
  • preinvasive disease;
  • molecular changes

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

The authors hypothesized that Lynch syndrome (LS)-associated endometrial cancer (EC) develops from morphologically normal endometrium that accumulates enough molecular changes to progress through a continuum of hyperplasia to carcinoma, similar to sporadic EC. The primary objective of the current study was to determine whether LS-associated EC involves progression through a preinvasive lesion. The secondary objective was to identify molecular changes that contribute to endometrial carcinogenesis in patients with LS.

METHODS

Women with a confirmed mismatch repair gene mutation for LS who were undergoing a prophylactic or therapeutic hysterectomy were eligible. Cases and controls were matched for EC and hyperplasia based preferentially on age and histology. Mutation status of phosphatidylinositol 3-kinase (PIK3CA); KRAS; AKT; LKB1; catenin (cadherin-associated protein), beta 1, 88kDa (CTNNB1); and phosphatase and tensin homolog (PTEN) protein loss was assessed.

RESULTS

Concurrent complex atypical hyperplasia (CAH) was found in EC in 11 cases of LS (39.3%) and 21 sporadic cases (46.6%). Loss of PTEN expression was common in both sporadic (69%) and LS-associated EC (86.2%). There was no significant difference noted with regard to the frequency of KRAS mutations in cases of sporadic EC (10.3%) compared with LS-associated EC (3.4%). AKT and LKB1 mutations were rarely observed. Mutations in PIK3CA and CTNNB1 occurred more frequently in cases of sporadic EC compared with LS-associated EC.

CONCLUSIONS

Hyperplasia, particularly CAH, is part of the preinvasive spectrum of disease in LS-associated EC, as indicated by the presence of complex hyperplasia and CAH in cases of LS. Although loss of PTEN is common in both LS and sporadic EC cases, there was a lack of additional mutations in LS-associated EC cases. This suggests that in the context of the mismatch repair defects in LS, fewer additional molecular changes are required to progress from preinvasive lesions to cancer. Cancer 2013;119:3027—3033. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Endometrial cancer (EC) is the most common gynecologic malignancy in the United States; approximately 5% of cases are attributed to an inherited predisposition.[1] Lynch syndrome (LS), previously known as hereditary nonpolyposis colorectal cancer, accounts for the majority of such cases. Women with LS have a 40% to 60% predicted lifetime risk of developing EC, in addition to a 40% to 60% lifetime risk of developing colorectal cancer.[2-4] The risk of EC appears to differ slightly based on the specific germline mutation in mismatch repair (MMR) genes MLH1 (MutL homolog 1, colon cancer, nonpolyposis type 2 [E. coli]), MSH2 (mutS homolog 2, colon cancer, nonpolyposis type 1 [E. coli]), and MSH6 (mutS homolog 6 [E. coli]). The estimated lifetime risk at age 70 years is 40% for MSH2 carriers, 27% for MLH1 mutation carriers, and 26% for MSH6 carriers.[5-7]

Similar to the well-established adenoma-to-carcinoma sequence of colon cancer development, sporadic EC is believed to develop through a continuum of complex atypical hyperplasia (CAH) to well-differentiated cancer.[8] The risk of endometrial hyperplasia progressing to cancer has been previously characterized in a large, nested, case-control study by Lacey et al.[9] This study demonstrated a cumulative progression risk of 4.6% for endometrial hyperplasia (simple or complex) without atypia to cancer over 20 years and a 30% cumulative risk for the progression of atypical hyperplasia to cancer.[9] Risk factors for sporadic endometrioid EC include age, obesity, nulliparity, and an excess of estrogen to progesterone.[10] In early carcinogenesis, the endometrium is believed to accumulate molecular changes that lead to CAH, the direct precursor lesion to endometrioid endometrial cancers.[11] These cancers are commonly associated with abnormalities in the tumor suppressor phosphatase and tensin homolog (PTEN), oncogenes KRAS and beta-catenin, and in the DNA MMR genes. For patients with LS-associated cancers, it is unknown whether the endometrium progress through a continuum of precancerous lesions to cancer similar to sporadic EC. Furthermore, although previous studies have shown that patients with LS-associated colorectal cancer progress in an accelerated fashion from preinvasive lesions to invasive disease, it is not known whether this accelerated progression is present in the development of EC.[12] The presence of endometrial hyperplasia in LS carriers has been reported, but to our knowledge few studies to date have attempted to identify important molecular changes associated with malignant transformation in this population.[11-13]

We hypothesized that LS-associated EC develops from morphologically normal endometrium that accumulates molecular changes to progress through a continuum of hyperplasia to carcinoma similar to sporadic EC. The primary objective of the current study was to determine whether LS-associated endometrial carcinogenesis involves progression through a preinvasive lesion. The secondary objective was to identify molecular changes that contribute to endometrial carcinogenesis in women with LS. To better understand the role of preinvasive lesions in LS-associated endometrial carcinogenesis, we examined molecular changes in preinvasive and invasive lesions in a cohort of women with LS who underwent hysterectomy for prophylaxis or for the treatment of endometrial cancer.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Study Participants

Institutional Review Board approval was obtained for the current study. Women with a confirmed MMR gene mutation (MLH1, MSH2, or MSH6) for LS who were undergoing a prophylactic or therapeutic hysterectomy were eligible. Cases were selected based on the availability of paraffin-embedded tissue in The University of Texas MD Anderson Cancer Center tissue bank as well as from patients who were enrolled in the LS registry and who had provided tumor blocks for analysis. Case controls were matched 2:1 for sporadic EC and CAH based preferentially on age and histology. Case controls were further chosen to approximate the year of diagnosis and stage of disease from the tissue archive at The University of Texas MD Anderson Cancer Center.

Clinical parameters were abstracted from patients' medical records. Patients were divided into 3 groups based on their histopathologic diagnoses: EC, hyperplasia, and benign endometrium.

Immunohistochemical Analysis and Pathology Review

Hematoxylin and eosin-stained slides for corresponding endometrial specimens were acquired for all patients and reviewed by a single gynecologic pathologist (B.D.) to confirm the diagnosis and to establish the presence or absence of concurrent endometrial hyperplasia in cases of EC. Pathologic information in cases of EC, including histology, grade, stage, and coexistence of endometrial hyperplasia, were ascertained. Expression of PTEN was assessed by immunohistochemistry (IHC) using a monoclonal mouse antihuman PTEN, clone 6H2.1 (Dako North America Inc, Carpinteria, Calif) at a 1:100 dilution. PTEN IHC for all samples was scored by 1 pathologist (B.D.) as positive, negative, or heterogeneous using an established scoring system.[14] In all cases, stromal cells and blood vessels had intensely positive expression for PTEN, thus serving as internal positive controls. Endometrial tissue that was considered to be positive demonstrated diffuse positive cytoplasmic and nuclear staining in the majority of cells (Fig. 1Top). Positive staining in endometrial tissue (cancer, CAH, or benign) was comparable to that detected in normal stromal cells. Endometrial tissue in which no or only rare cells stained was considered negative for PTEN (< 10%) (Fig. 1Bottom). Endometrial tissue with distinct areas of positive and negative staining was designated as having a heterogeneous staining pattern, which indicates local loss of PTEN. Cases with heterogeneous staining were combined with those staining negative for PTEN for the purposes of analysis. Previous studies have shown this immunohistochemical method is superior to PTEN sequencing for determining PTEN protein loss.[14]

image

Figure 1. (Top) Endometrial tissue considered to be positive demonstrated diffuse positive cytoplasmic and nuclear staining in the majority of cells. (Bottom) Endometrial tissue with no or only rare cells staining was considered negative for phosphatase and tensin homolog (PTEN).

Download figure to PowerPoint

Mutational Analysis

Tumor DNA was isolated according to standard procedures. Mutational analysis was performed using the Sequenom MassARRAY analyzer (Sequenom Inc, San Diego, Calif) as described previously.[15] Briefly, a mass spectroscopy-based approach evaluating single nucleotide polymorphisms was used to detect hotspot mutations in our genes of interest (PIK3CA, KRAS, AKT, LKB1, and CTNNB1). Polymerase chain reaction and extension primers were designed using an assay design from Sequenom.

Statistical Analysis

Statistical analysis was performed using the Fisher exact test, chi-square test, and 1-way analysis of variance. One-way analysis of variance was used to compare differences in age among LS carriers. A P value < .05 was considered statistically significant for all tests.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

There were 67 patients with MLH1, MSH2, or MSH6 mutations in the LS cohort and 84 case-matched controls (hyperplasia and sporadic EC cases were matched 2:1 for each LS case) enrolled in the study. Clinical characteristics are summarized in Table 1. Among patients with LS, 30 patients had normal endometrium, 8 had hyperplasia (2 with complex hyperplasia without atypia and 6 with CAH), and 29 patients had EC. In the control group, there were 10 patients with normal endometrium, 16 with CAH, and 58 patients with EC. In the LS cohort, the median age at the time of surgery was 43 years (range, 32 years-57 years) in the group with normal endometrium, 39 years (range, 35 years-49 years) in the group with hyperplasia, and 49 years (range, 37 years-62 years) in the group with EC. Women with LS-associated EC were significantly older than both LS carriers with normal endometrium (P = .0025) and women with hyperplasia (P = .0019). Mutations in MSH2 were the most common MMR mutation noted in LS carriers. Women with sporadic EC were significantly more obese (P = .008) compared with women with LS-associated EC. There was no significant difference observed between women in the 3 groups with regard to parity (P = .33).

Table 1. Clinical Characteristics
 Lynch SyndromeCase-Matched Cohort
EC (n= 29)Hyperplasia (n = 8)aBenign (n= 30)EC (n=58)CAH (n=16)Benign (n=10)
  1. Abbreviations: BMI, body mass index; CAH, complex atypical hyperplasia; EC, endometrial cancer; MLH1, MutL homolog 1, colon cancer, nonpolyposis type 2 (E. coli); MSH2, mutS homolog 2, colon cancer, nonpolyposis type 1 (E. coli); MSH6, mutS homolog 6 (E. coli); NA, not applicable.

  2. a

    Includes complex hyperplasia without atypia (n=2) and complex hyperplasia with atypia (n=6).

  3. b

    P < .05.

  4. c

    P = .008.

Median age, (range), y      
 49 (37-62)b39 (35-49)43 (32-57)52 (31-61)52 (40-61)51 (25-54)
Mutation      
MLH18 (27.6%)3 (37.5%)11 (36.7%)NANANA
MSH216 (55.2%)5 (62.5%)17 (56.7%)   
MSH65 (17.2%)02 (6.6%)   
BMI      
≤24.912 (42.9%)6 (75%)13 (48.2%)5 (8.6%)6 (37.5%)4 (40%)
25-29.97 (25%)06 (22.2%)11 (19%)2 (12.5%)1 (10%)
≥309 (32.1%)2 (25%)8 (29.6%)42 (72.4%)c8 (50%)5 (50%)
       
Parity      
09 (31%)06 (20%)22 (38%)6 (40%)4 (40%)
≥120 (69%)8 (100%)24 (80%)36 (62%)9 (60%)6 (60%)
       

Pathologic characteristics of cases of LS-associated EC and case-matched controls are summarized in Table 2. Only 17.9% of LS-associated EC were grade 1 (Grade 1 is 95% or more of the cancerous tissue forming glands) tumors, 32.1% were grade 2 (Grade 2 is between 50% and 94% of the cancerous tissue forming glands) tumors, and 50% were grade 3 (Grade 3 is less than half of the cancerous tissue forming glands) tumors. Among the sporadic EC cases, 12.1% were grade 1 tumors, 74.1% were grade 2 tumors, and 13.8% were grade 3 tumors. The majority of both LS-associated and case-matched EC cases were stage I (The International Federation of Obstetrics and Gynecology (FIGO) 2009) at the time of diagnosis (75.9% and 67.2%, respectively). In LS-associated EC, the majority (62.1%) of the ECs were of endometrioid histology, and another 24.1% were of a mixed histology.

Table 2. Pathologic Characteristics of Endometrial Cancer
 Lynch Syndrome EC (n=29)Case-Matched Cohort EC (n=58)
  1. Abbreviations: EC, endometrial cancer; UPSC, uterine papillary serous carcinoma.

Grade  
15 (17.2%)7 (12.1%)
210 (34.4%)43 (74.1%)
314 (50%)8 (13.8%)
Stage of disease  
I22 (75.9%)39 (67.2%)
II3 (10.3%)7 (12.1%)
III/IV4 (13.8%)12 (20.7%)
Histology  
Endometrioid18 (62.1%)54 (93.1%)
Mixed7 (24.1%)3 (5.2%)
Undifferentiated2 (6.9%)0
UPSC2 (6.9%)1 (1.7%)

Table 3 displays the frequency of EC with concurrent hyperplasia in LS and the case-matched cohort. Confirming the importance of complex hyperplasia in the progression to EC, concurrent hyperplasia was common in both LS and case-matched EC cases. In LS-associated EC, 11 cases (37.9%) had concurrent CAH, in addition to 1 case (3.4%) with concurrent CH without atypia and 1 case (3.4%) with simple hyperplasia without atypia. In the case-matched cohort, 21 cases (36.2%) displayed concurrent CAH. None of the women in either the LS or sporadic cohort with EC arising in a background of hyperplasia was found to have an advanced stage of disease (stage III or stage IV).

Table 3. Endometrial Cancers With Concurrent Hyperplasia
CaseLynch SyndromeCase-Matched Cohort
  1. Abbreviations: CAH, complex atypical hyperplasia; CH, complex hyperplasia; EC, endometrial cancer; SH, simple hyperplasia.

Total no. of EC cases2958
EC with hyperplasia13 (44.8%)21 (36.2%)
Concurrent SH without atypia1 (3.4%)0
Concurrent CH without atypia1 (3.4%)0
Concurrent CAH11 (37.9%)21 (36.2%)

Common gene mutations previously reported in sporadic EC (PTEN, PIK3CA, KRAS, and CTNNB1) were also found to be present in our sporadic EC cohort and, to a lesser extent, in the LS-associated EC cohort. As shown in Table 4, in this set of LS-associated EC cases, there were 4 PIK3CA mutations (13.8%), 1 KRAS mutation (3.4%), and 2 CTNNB1 mutations (6.9%). In a similar fashion, in the sporadic EC cases, there were 23 PIK3CA mutations (39.7%), 5 KRAS mutations (8.6%), 18 CTNNB1 mutations (31%), 1 LKB1 mutation (1.7%), and 1 AKT mutation (1.7%). There was a similar frequency of KRAS mutations in both sporadic EC and LS-associated EC cases. However, CTNNB1 and PIK3CA mutations both occurred significantly more frequently in the cohort with sporadic EC compared with the LS-associated EC cohort (P = .002 and P = .015, respectively).

Table 4. Mutational Analysis
 Lynch SyndromeSporadic
 EC (n=29)Hyperplasia (n=8)Benign (n=30)EC (n=58)CAH (n=16)Benign (n=10)
  1. Abbreviations: CAH, complex atypical hyperplasia; CTNNB1, catenin (cadherin-associated protein), beta 1, 88kDa; EC, endometrial cancer; PIK3CA, phosphatidylinositol 3-kinase.

  2. a

    P < .05 compared with Lynch syndrome-associated EC.

PIK3CA4 (13.8%)01 (3.3%)23 (39.7%)a1 (6.3%)0
KRAS1 (3.4%)005 (8.6%)00
CTNNB12 (6.9%)01 (3.3%)18 (31.0%)a01 (10%)
LKB10001 (1.7%)00
AKT0001 (1.7%)01 (10%)

In both LS-associated and sporadic EC cases, PTEN loss by IHC was a common occurrence (86.2% and 69.0%, respectively) (Table 5). Among the hyperplasia cases, only 1 of 6 LS-associated CAH (12.5%) demonstrated PTEN loss whereas 11 of 16 of the sporadic CAH cases s (68.8%) did so. Furthermore, PTEN loss was more common in benign endometrium in patients with LS compared with average-risk women (43.3% vs 10%; P = .057). Coaberrations in PTEN/PIK3CA or PTEN/PIK3CA/KRAS were less common in LS-associated endometrioid EC compared with case-matched endometrioid EC (Table 6). In sporadic EC cases with concurrent CAH, both PIK3CA mutations and PTEN loss were common. However, in LS-associated EC cases with concurrent CAH, only PTEN loss was regularly observed.

Table 5. PTEN Status Determined by Immunohistochemistry
 Lynch Syndrome 
  Hyperplasia Case-Matched Cohort
 EC (n=29)CAH (n=6)CH (n=2)Benign (n=30)EC (n=58)CAH (n=16)Benign (n=10)
  1. Abbreviations: CAH, complex atypical hyperplasia; CH, complex hyperplasia; EC, endometrial cancer; PTEN, phosphatase and tensin homolog.

PTEN loss25 (86.2%)1 (16.7%)013 (43.3%)40 (69.0%)11 (68.8%)1 (10%)
PTEN present4 (13.8%)5 (83.3%)2 (100%)17 (56.7%)18 (31.0%)5 (31.2%)9 (90%)
Table 6. Coaberrations
Abnormal PathwayLynch Syndrome EC (n=29)Case-Matched EC (n=58)
EndometrioidNonendometrioidEndometrioidNonendometrioid
  1. Abbreviations: EC, endometrial cancer; PIK3CA, phosphatidylinositol 3-kinase; PTEN, phosphatase and tensin homolog.

PTEN loss plus PIK3CA mutation only22111
KRAS mutation (activation) only0141
PTEN loss plus PIK3CA mutation plus KRAS mutation (activation)0151

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Sporadic endometrioid EC is believed to develop through a continuum of CAH to well-differentiated cancer.[8] Bokhman previously reported that an estimated 30% to 40% of sporadic endometrioid EC cases coexist with CAH.[16] Furthermore, as demonstrated by Lacey et al, patients with hyperplasia without atypia and CAH had a progressive increase in their lifetime risk of progression to EC.[9] In a screening study of LS patients by Renkonen-Sinisalo et al, the authors demonstrated the presence of endometrial hyperplasia among asymptomatic women during surveillance endometrial biopsy in 8% of Lynch mutation carriers (both complex hyperplasia and CAH were observed).[17] In our cohort of patients with LS-associated EC, we found that complex hyperplasia and CAH existed in 12% of the LS carriers. CAH also occurred concurrently with EC in 37.9% of LS cases, thereby supporting the likelihood that CAH is part of the preinvasive disease spectrum. LS-associated EC with concurrent CAH was found to occur primarily in the endometrioid histotype (9 of 11 cases). These findings of endometrial hyperplasia in Lynch mutation carriers lend further support to the preinvasive disease continuum that may include complex hyperplasia with or without atypia.

To the best of our knowledge, very few studies to date have reported on molecular alterations in hyperplasia and EC from LS carriers. In a study with 41 cases, Zhou et al attempted to determine whether PTEN is involved in the pathogenesis of EC in Lynch mutation carriers and whether PTEN inactivation precedes MMR deficiency.[18] The authors demonstrated a 68% PTEN loss by IHC in their patients with LS-associated EC. They then performed mutation analysis on 20 of the 28 PTEN null cases and found that 17 of the 20 cases (85%) had a somatic PTEN mutation.[18] PTEN loss occurred in approximately 86% of cases of LS-associated EC and was observed in 43% of LS benign endometrium samples. Future studies with larger numbers of patients with LS-associated CAH will help determine whether PTEN is an early event in the development of LS-associated EC. In the population in the current study, PIK3CA mutations were found to be more common in cases of sporadic EC, and KRAS and CTNNB1 mutations were noted infrequently in LS carriers. However, KRAS and CTNNB1 occurred in sporadic EC cases in the current study at a frequency similar to previously reported data. The lack of other mutations in LS-associated EC cases in the current study may suggest that in the presence of an existing MMR deficiency, loss of PTEN function may be sufficient to drive tumorigenesis. Future studies in which CAH can be microdissected from concomitant EC may be helpful to delineate the order of mutations in patients with LS-associated EC pathogenesis. Cohn et al examined CAH occurring concomitantly with EC but physically remote from the cancer for microsatellite instability and KRAS mutations (but not tested specifically for MMR mutations). They observed that some cases of microsatellite instability-positive CAH specimens lacked the KRAS mutation noted in the coexisting cancer, thereby supporting their proposed model in which loss of DNA MMR precedes KRAS mutation.[19] This suggests that although LS-associated EC appears to progress along the continuum of CAH to cancer in a manner that is similar to sporadic EC, the molecular changes accumulated during this progression may be different in patients with LS.[20] It is believed that women with LS develop DNA MMR abnormalities, resulting in decreased DNA repair and an increased loss of PTEN. In our cohort of cases with LS, the absence of additional mutations suggests that fewer molecular changes are necessary to drive tumorigenesis. Future studies with more comprehensive mutational analysis will be essential to determine whether LS-associated EC involves overall fewer genetic mutations, which specific gene mutations are involved, and which gene mutations occur in CAH versus EC.

Other known risk factors in patients with sporadic EC have not to our knowledge been well documented in LS mutation carriers. Chronic unopposed estrogen exposure is believed to trigger the transformation to hyperplasia followed by EC. In the current study, parity did not appear to differ significantly across groups. However, obesity was significantly more frequent in women with sporadic EC, indicating perhaps a less-important role of obesity in LS mutation carriers. It is unclear in women with LS whether obesity and superfluous estrogen have an additive effect; future investigation is warranted to elucidate this finding.

One limitation of the current study is that it is a retrospective review with an inherent ascertainment bias. All patients were self-selected or approached because they were LS mutation carriers. It is interesting to note that the population in the current study is different from those in other large studies of LS mutation carriers in that more than one-half of the women in the current study were MSH2 carriers. This differs from the majority of European study populations from Scandinavian registries, in which mutations are predominantly in the MLH1 gene.[17, 21] This may have broader implications for the type of mutations that are found in the US population, because MSH2 mutations carry an overall increased lifetime risk of developing EC.

The results of the current study indicate that hyperplasia, particularly CAH, is part of the preinvasive spectrum of disease in patients with LS-associated EC, as indicated by the presence of complex hyperplasia and CAH in women with LS. Although PTEN loss is common in both LS and sporadic EC cases, there was a lack of PIK3CA and CTNNB1 mutations in LS-associated EC cases. Additional studies with more cases are needed to identify molecular mutations in the LS-associated CAH to help elucidate the sequence of alterations leading to tumorigenesis and whether this progression is accelerated in EC. Furthermore, knowledge of the molecular alterations can facilitate potential targeted therapeutic interventions in the treatment of LS-associated EC.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Research reported in this publication was supported by the National Institute of Health Ruth L. Kirschstein National Research Service Award (NRSA) under Award Number T32 CA101642. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The project described was supported in part by Grant Number P50CA098258 from the National Cancer Institute.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES