SEARCH

SEARCH BY CITATION

Keywords:

  • BRCA1 gene;
  • BRCA2 gene;
  • polymerase chain reaction;
  • mutation analysis;
  • genetic counseling;
  • breast cancer risk

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Factors that are predictive of carrying BRCA1 and BRCA2 germline mutations in patients with breast carcinoma are awaited widely. The genetically homogeneous Sardinian population may be useful for defining the role of such genetic alterations further through a clinical evaluation program.

METHODS

One hundred two of 659 patients with breast carcinoma (15.5%) who were collected consecutively had a family history of breast carcinoma and were screened for BRCA1/2 mutations by denaturing high-performance liquid chromatography and DNA sequencing.

RESULTS

Three deleterious germline BRCA1/2 mutations were detected in 15 of 102 families (14.7%), including 13 families (86.7%) with BRCA2 mutations and 2 families (13.3%) with BRCA1 mutations. A single variant, BRCA2-8765delAG, was the most recurrent mutation in the series and was found in 12 of 102 families (11.8%) and in 18 of 657 patients (2.7%). The average age at diagnosis was significantly younger in families with BRCA1/2 mutations (48.6 yrs) compared with the age of patients who had no detectable mutation (52.9 yrs; P = 0.039). Moreover, BRCA1/2 mutations were found at a significantly higher rate in families who had at least 1 member with ovarian carcinoma or male breast carcinoma (5 of 12 families; 41.7%) than in families without such an association (10 of 90 families; 11.1%; P = 0.003).

CONCLUSIONS

BRCA2 mutations were approximately 6 times more prevalent than BRCA1 mutations. A diagnosis of breast carcinoma before age 50 years, ovarian carcinoma, male breast carcinoma, and 3 affected generations all were associated significantly with BRCA1/2 mutations. Although the current findings provided further support for the hypothesis that additional breast carcinoma susceptibility genes remain to be identified, such indicators of the presence of BRCA1/2 mutations may be useful in counseling patients about undergoing genetic testing. Cancer 2005. © 2005 American Cancer Society.

Germline mutations in 2 tumor suppressor genes, BRCA1 (Mendelian Inheritance in Man [MIM] 113705) and BRCA2 (MIM 600185), are associated with hereditary susceptibility to breast and ovarian carcinoma.1 Overall, it is estimated that the penetrance of BRCA1 mutations is 50–80% for breast carcinoma and 20–50% for ovarian carcinoma2; for BRCA2 mutations, the estimated penetrance is 30–80% for breast carcinoma and 10–20% for ovarian carcinoma by age 70 years.3 It has been demonstrated that BRCA1/2 mutation carriers also present with a higher risk for other types of carcinoma, such as prostate carcinoma and malignant melanoma.2, 4

According to the Breast Cancer Information Core data base,5 the majority of germline alterations identified in BRCA1 and BRCA2 is unique (57% and 63%, respectively); the remaining mutations are recurrent founder mutations that have been described in different ethnic groups and populations.6–21 Among recurrent BRCA1/2 mutations, some are population-specific, whereas others occur in DNA sequences that may represent mutational hot-spot regions.22–24 The proportion of recurrent mutations to unique mutations varies in different populations and subpopulations, reflecting historic influences on migration, population structure, and geographic and cultural isolation.23 Such differences in mutation prevalence clearly indicate that there is a need to define the risk of breast cancer in population-based studies as well as both the type and frequency of BRCA1/2 mutations within every cohort or country.

In Sardinia, which has experienced little immigration due to its remote location and has a population that has inherited many of the same genetic traits, the contribution of BRCA1/2 mutations to the population incidence of breast carcinoma has not been estimated carefully to date. Breast carcinoma represents the principal death-causing malignancy in Sardinia, with an incidence (standardized rate, 95 per 100,000 inhabitants per yr; Sardinian population includes approximately 1.5 million inhabitants) quite comparable to that observed in Western countries.25, 26 The median age of onset for breast carcinoma among Sardinian women is 65 years.

In recent years, numerous studies have assessed the BRCA1/2 mutation prevalence in various cohorts, although few have evaluated the predictors for the occurrence of both BRCA1 and BRCA2 mutations in a clinic-based population. In this report, we provide such an analysis of 102 patients with breast carcinoma who presented with high-risk features due to their familial recurrence of the disease from a series of 659 consecutively collected clinical patients. Our data are important in defining further the likelihood of identifying BRCA1 and BRCA2 mutations in patients who present to high-risk cancer evaluation clinics to provide accurate guidance to women and families about considering genetic testing. Moreover, in the current study, we assessed the real contribution of BRCA1/2 mutations to breast carcinoma predisposition in the Sardinian population.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patient Population

Six hundred fifty-nine patients with breast carcinoma who originated from North Sardinia were recruited from clinics at the University of Sassari and Azienda Unitá Sanitaria Locale 1 of Sassari (which represent the principal institutions accounting for patients with cancer from the Central and Northern parts of the island). Sardinian origin was ascertained in all patients through genealogic studies. Patients with a histologically proven diagnosis of breast carcinoma were collected consecutively during the previous 4 years; no additional selection criteria were used for their inclusion into the study.

Family history of cancer was evaluated through specific questionnaires during the follow-up visits at the different departments of the participating institutions. One hundred two familial cases were identified. Families were eligible if at least three affected members (including the proband and first-degree or second-degree relatives) were present. All information was verified through careful analysis of the hospital records; the cancer diagnosis of each affected family member was confirmed by pathology reports. In Table 1, families are categorized according to the total number of patients with breast carcinoma, the association with prostate or ovarian carcinoma or with male breast carcinoma, and the distribution of patients through generations.

Table 1. Characteristics of the Breast Cancer Families
CharacteristicNo. of families
Familial recurrence 
 Breast site-specific carcinoma90
 Breast and ovarian carcinoma8
 Male breast carcinoma4
 Total102
No. of family members with breast carcinoma 
 354
 419
 > 529
Family history features 
 Mother with breast carcinoma27
 Sister(s) with breast carcinoma21
 Both mother and sister(s) with breast carcinoma10
 Other relatives with breast carcinoma44
 Three affected generations18
 Two affected generations84
Age at diagnosis of family probands 
 < 45 yrs36
 46–59 yrs40
 > 60 yrs26
 Median age (yrs)47
 Age range (yrs)25–80

A written informed-consent form was obtained for both genetic counseling and blood/tissue sampling before genetic testing. Germline DNA was isolated from 1) peripheral blood samples of at least one affected member (proband) in each family and 2) peripheral blood or paraffin embedded normal tissues among the remaining nonfamilial patients with breast carcinoma. The study was reviewed and approved by ethical review boards at both institutions (University and Azienda Unitá Sanitaria Locale 1 of Sassari).

Mutation Analysis

All patients were analyzed for germline mutations throughout the entire coding sequences and intron-exon boundaries of the BRCA1 and BRCA2 genes. Mutation screening was performed by denaturing high-performance liquid chromatography (DHPLC) followed by automated sequencing. Protocols for polymerase chain reaction (PCR)-based amplification and mutation analysis of exons and exon-intron boundaries were reported previously.23 DHPLC analysis was performed with the Wave® nucleic acid fragment-analysis system (Transgenomic, Santa Clara, CA). Suspected variants were visualized as a characteristic pattern of peaks corresponding to the mixture of homoduplex and heteroduplex formed when wild type and mutant DNA were hybridized. Abnormal PCR products identified by DHPLC analysis were sequenced directly using an automated fluorescence-cycle sequencer (ABI PRISM 3100; Applied Biosystems, Foster City, CA).

Statistical Analysis

Univariate analysis for correlations between genetic markers (mutations in BRCA1, BRCA2, or either gene) and baseline variables (family and proband characteristics) was carried out by using the chi-square test. Characteristics included the following: the presence of ovarian or male breast carcinoma, the number of affected family members (three or four vs. five or more), the number of affected generations (two vs. three or more), age at diagnosis of all family members with breast carcinoma,; pathologic tumor extension (pT) according to the TNM classification,28 and the age of onset in the proband. Odds ratios of carrying BRCA1/2 mutations were estimated by using a logistic regression model and are reported with 95% confidence intervals (95%CI). Except for the age at diagnosis and the pT extension values, which were analyzed as continuous variables, the remaining features were analyzed as dichotomous variables (presence vs. absence).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Among the 659 consecutively collected clinical patients, 102 patients (15.5%) had a family history of breast carcinoma. All families had at least three individuals diagnosed with breast carcinoma, with an average of four diagnoses of breast carcinoma per family (median, three diagnoses; range, from three to nine diagnoses). The eight families that had a history of both breast and ovarian carcinoma had an average of two diagnoses of ovarian carcinoma (median, one diagnosis; range, from one to seven diagnoses). Four families of female probands had at least one diagnosis of male breast carcinoma (Table 1).

Mutation screening for all coding regions and splice boundaries of BRCA1 and BRCA2 genes was performed by DHPLC analysis on germline DNA from family probands; all PCR products that had an abnormal denaturing profile compared with the normal controls were sequenced using an automated approach. Among the 139 probands from the 102 families studied, 3 germline coding region mutations of known functional significance were detected in either BRCA1 or BRCA2 among 15 families (14.7%) (Table 2). All 3 mutations were absent in normal genomic DNA from 103 unrelated, healthy individuals (corresponding to 206 control chromosomes) and were classified as disease-causing variants due to their predicted effects on proteins. Most of these mutations (13 mutations; 86.7%) were found in BRCA2, and only small numbers (2 mutations; 13.3%) were found in BRCA1. The mutation BRCA2-8765delAG, which we described previously as a frameshift mutation with founder effect in North Sardinia,29 remained the only recurrent mutation among families from our series (prevalence, 12 of 102 families; 11.8%) (Table 2). The distribution of familial members carrying the BRCA2-8765delAG variation and originating from different villages in North Sardinia further confirmed the founder effect of this germline mutation in such a geographic area (Fig. 1). Families with and without mutations had the same percentage of women affected with breast carcinoma (27% and 26% of women age older than 21 yrs, respectively).

Table 2. BRCA1 and BRCA2 Mutations Identified by Screening 102 Families with Breast Cancer
No. of positive families (%)GeneExonNucleotideCodonBase changeAmino acid changeMutation effectMutation designation
2 (1.96)BRCA1111714505A to TLys to StopNonsenseLys505ter
12 (11.8)BRCA22087652845delAGStop 2867Frameshift8765delAG
1 (0.98)BRCA21160241943delTAStop 6024Frameshift6024delTA
thumbnail image

Figure 1. This map illustrates the distribution of patients with the familial BRCA2-8765delAG germline mutation in North Sardinia. Geographic areas of the island from which patients were collected are indicated (the two shades correspond to different provinces). Dots indicate the origin of patients who presented with the BRCA2-8765delAG mutation.

Download figure to PowerPoint

To evaluate the prevalence of the BRCA2-8765delAG founder mutation in North Sardinia, we screened the remaining 555 genomic DNA samples from nonfamilial breast carcinoma patients (for 2 patients, we were not able to perform the mutation analysis). The BRCA2-8765delAG mutation was found in 6 patients (1.1%) who originated from different villages. Overall, such a founder mutation was observed in 18 of 657 consecutively collected Sardinian breast carcinoma patients (2.7%). Moreover, no additional individual who carried either the BRCA1-Lys505ter or the BRCA2-6024delTA mutation was detected after screening of blood DNA samples from individuals with nonfamilial breast carcinoma in our series (altogether, such germline mutations were observed in 2 of 657 [0.3%] and 1 of 657 [0.15%] consecutively collected Sardinian patients with breast carcinoma, respectively).

Table 3 shows that the proportion of families carrying BRCA1/2 germline mutations differed strikingly when we analyzed several family or proband phenotypes. Twelve families had at least 1 case of ovarian carcinoma or at least 1 case of male breast carcinoma, of which 5 families (41.7%) had either a BRCA1 or BRCA2 mutation (Table 3). The remaining families that did not present such an association had a much lower frequency of BRCA1/2 mutations (10 of 90 families; 11.1%). The BRCA1/2 mutations were identified in 5 of 20 patients with synchronous or asynchronous bilateral tumors (25%) and in 10 of 82 patients with unilateral breast carcinoma (12%) (Table 3).

Table 3. Distribution of BRCA1/BRCA2 Mutations among Breast Cancer Families
CharacteristicNo. of familiesBRCA1/2-positive families (%)BRCA1/2-negative families (%)
  • a

    P = 0.211 (2-tailed chi-square test; 95% confidence interval).

  • b

    P = 0.039 (2-tailed chi-square test; 95% confidence interval).

Total10215 (14.7)87 (85.3)
Familial recurrence   
 Families with breast site-specific carcinoma9010 (11.1)80 (88.9)
 Families with breast and ovarian carcinoma83 (37.5)5 (62.5)
 Families with female and male breast carcinoma42 (50.0)2 (50.0)
No. of family members with breast carcinoma   
 < 5737 (9.6)66 (90.4)
 > 5298 (27.6)21 (72.4)
Family history features   
 Mother with breast carcinoma271 (3.7)26 (96.3)
 Sister(s) with breast carcinoma214 (19.0)17 (81.0)
 Mother and sister(s) with breast carcinoma103 (30.0)7 (70.0)
Other relatives with breast carcinoma447 (15.9)37 (84.1)
Three affected generations186 (33.3)12 (66.7)
Two affected generations849 (10.7)75 (89.3)
Age at diagnosis of family probands   
 < 45 yrs366 (16.7)30 (83.3)
 46–59 yrs405 (12.5)35 (87.5)
 > 60 yrs266 (23.1)20 (76.9)
Proband breast carcinoma focality   
Unilateral primary tumor8210 (12.2)72 (87.8)
Bilateral (synchronous or asynchronous) primary tumors205 (25.0)15 (75.0)
Proband average age at diagnosis (yrs)50.7 (51.9%)a  
Family average age at diagnosis (yrs)48.6 (52.9%)b  

The average age at the time of breast carcinoma diagnosis in female probands and in all women affected with breast carcinoma in each family was compared on the basis of the mutation status (Table 3). The average age of onset was significantly younger in families that carried BRCA1/2 mutations than in families with no detectable mutation (48.6 yrs vs. 52.9 yrs, respectively; P = 0.039) (Table 3). No significant difference was observed in the average age of probands from our series (patients who were positive or negative for BRCA1/2 mutations were diagnosed with breast carcinoma at an average age of 50.7 yrs and 51.9 yrs, respectively; P = 0.21) (Table 3).

Statistical analysis was performed to assess for the presence of any significant association between mutations in BRCA1, BRCA2, or either gene and phenotype characteristics (the presence of ovarian or male breast carcinoma, the number of affected members, the number of affected generations in the family, the presence of synchronous or asynchronous bilateral breast carcinoma, primary tumor extension, and age at diagnosis in the probands). Table 4 shows the results from the univariate analysis, which indicate that earlier age at diagnosis (P = 0.005), the presence in the family of ovarian or male breast carcinoma (P = 0.003), ≥ 5 women with breast carcinoma (P = 0.046), and 3 affected generations (P = 0.019) were correlated significantly with the presence of a BRCA1/2 mutation. No other family or proband feature was correlated statistically with predisposing BRCA1/2 mutations. After multivariate analysis, an association with ovarian or male breast carcinoma (P = 0.011) and the number of affected generations (P = 0.014) remained statistically independent predictive factors for the occurrence of BRCA1/2 germline mutations (Table 5).

Table 4. Univariate Analysis between Proband or Family Features and BRCA1/BRCA2 Germline Mutations
Proband's characteristicsOR95%CIP valuea
  • OR: odds ratio; 95%CI: 95% confidence interval.

  • a

    Two-tailed chi-square tests were used to determine P values and 95%CIs.

Age at diagnosis7.431.82–30.240.005
≥ One family member with ovarian or male breast carcinoma10.522.20–50.350.003
≥ One family member with prostate carcinoma0.260.04–1.620.149
No. of affected family members3.651.02–13.130.046
No. of affected generations5.291.31–21.420.019
Table 5. Multivariate Analysis between Proband or Family Features and BRCA1/BRCA2 Germline Mutations
Proband's characteristicsOR95%CIP valuea
  • OR: odds ratio; 95%CI: 95% confidence interval; pT: pathologic tumor extension.

  • a

    Two-tailed chi-square tests were used to determine P values and 95%CIs.

  • b

    pT was determined according to the TNM classification system.

Age at diagnosis1.010.95–1.060.684
Bilateral primary tumor(s)1.620.38–6.850.507
Primary tumor extension (pT)b0.360.02–4.010.242
≥ One family member with ovarian or male breast carcinoma9.531.65–54.900.011
No. of affected family members1.370.98–1.920.064
No. of affected generations6.021.43–25.100.014

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The identification of BRCA1 and BRCA2 mutation carriers represents an important step toward prevention and early detection of breast and ovarian carcinoma. Due to environmental and genetic factors, the prevalence of BRCA1/2 mutations is variable among different populations. A real estimation of the proportion of positive and negative tests that may be expected in a referral risk evaluation clinic, therefore, is fundamental to provide clinical recommendations for BRCA1/2 genetic tests that affect patient decisions and cost-effectiveness estimates.

In our analysis, 15% of families that sought genetic testing from our high-risk breast cancer clinic had a detectable mutation in BRCA1 or BRCA2. In addition, considering that there may have been some lack of sensitivity of our methods for mutation analysis with a subsequent decreasing number of positive results, the prevalence of BRCA1/2 mutations in Sardinian families remained low (although the Finnish population showed a similar prevalence11). This low mutation frequency found in our study may be explained by the fact that either the majority of our families (73 of 102 families; 72%) contained 3 or 4 members with female breast carcinoma or a very small fraction (12 of 102 families; 12%) was associated with ovarian and male breast carcinoma. In line with this hypothesis, a significant association with BRCA1/2 mutations was observed among families with at ≥ 5 members who had breast carcinoma or had an association with ovarian/male breast carcinoma (see Tables 4, 5). Although referral for cancer risk evaluation may select for a reasonable percentage of mutation carriers when proband and family characteristics are defined well, our data highlight the fact that the majority of women who will be evaluated at regional and national referral centers will not have detectable mutations.

In the current study, BRCA2 mutations were approximately 6 times more common that BRCA1 mutations. The ratio of BRCA1 mutations to BRCA2 mutations varies widely between populations globally: Icelandic breast carcinoma families have almost exclusively BRCA2 mutations,10 there are roughly equal numbers of BRCA1 and BRCA2 mutations in French-Canadian breast carcinoma families30 and in British women with early-onset breast carcinoma.31 In the United States, different studies have demonstrated a clear prevalence of BRCA1 mutations.32, 33 In Table 6, the frequencies of BRCA1 and BRCA2 germline mutations among different populations are reported.

Table 6. Incidence of BRCA1/BRCA2 Germline Mutations in Unselected Patients with Breast Carcinoma from Different Populations
Authors and referencesPopulationNo. of analyzed patientsFrequency of BRCA1/2 mutations (%)
BRCA1   
 Malone et al., 200033United States (white)1936.2
 Krainer et al, 199734United States (white)7312.0
 Hopper et al., 199935Australian3882.3
 Loman et al., 200113Swedish2346.8
 Anglian Breast Cancer Group, 200036British1,2200.7
 Papelard et al., 200037Dutch6422.1
 Syrjäkoski et al., 200038Finnish1,0350.4
 Van der Looij et al., 200021Hungarian5003.4
BRCA2   
 Krainer et al., 199734United States (white)732.7
 Hopper et al., 199935Australian3882.3
 Loman et al., 200113Swedish2342.1
 Anglian Breast Cancer Group, 200036British1,4351.3
 Syrjäkoski et al., 200038Finnish1,0351.6
 Van der Looij et al., 200021Hungarian5000.2
 Thorlacius et al., 199739Icelandic63210.1
 Thorlacius et al., 199840Icelandic57512.0
BRCA1/2   
 Robson et al., 199741Ashkenazi Jews23625.0
 Tonin et al., 200142French Canadian6113.0

Altogether, this variability may be due either to differences in the penetrance of BRCA1 and BRCA2 mutations or to population founder effects, which clearly is the case in Iceland as well as in our series. Indeed, it was demonstrated that BRCA2-8765delAG is a disease-causing mutation with founder effect in North Sardinia and remained the only recurrent mutation among families from such a geographic area. Overall, 18 of 657 patients (2.7%) with breast carcinoma who were collected consecutively from North Sardinia presented with the BRCA2-8765delAG variant, further confirming its role as a founder mutation.

The BRCA2-8765delAG variant also has been described as a founder mutation in French-Canadian families.30 Haplotype analysis of some Sardinian and French-Canadian families, as we reported previously,29 demonstrated that the BRCA2-8765delAG mutation is associated with different haplotypes in 2 such populations (DNA samples of French-Canadian BRCA2-8765delAG carriers kindly were provided by Dr. Tonin). These results support the hypothesis that the AG-rich sequence in which the BRCA2-8765delAG mutation occurs may be a mutational hot-spot.

Although it has been estimated that the percentage of hereditary breast carcinoma in Sardinia is comparable to that observed in other genetically homogeneous populations,25, 27 the prevalence of our BRCA2 founder mutation was lower than expected. Three BRCA1 founder mutations account for the vast majority of breast carcinoma families among Ashkenazi Jews,22 and a single BRCA2 mutation accounts for nearly all breast-ovarian carcinoma families in Iceland.10

Conflicting data have been reported about the role of BRCA1/2 mutations in families with an association between breast carcinoma and prostate carcinoma.43–45 From data the presented in the current study, BRCA1 and BRCA2 mutations seem to have little or no role in predisposition to carcinoma; 2 families (6%) were positive for BRCA1/2 germline mutations among 36 genomic DNA samples from families with recurrent breast and prostate carcinomas. Conversely, the association between breast carcinoma and ovarian carcinoma or male breast carcinoma was the most important high-risk factor for predicting the presence of BRCA1/2 mutations. Male breast carcinoma is a characteristic phenotype determined by alteration of the BRCA2 gene, with mutation frequencies reported in the range of 4–21%. Founder mutations contribute to higher prevalence rates of male breast carcinoma among patients from Iceland,39, 40 Sweden,13 Hungary,21 and the Ashkenazi Jewish population.6, 7, 41 Generally, men with breast carcinoma present with more advanced disease and a poorer prognosis,46 and assessment of their BRCA2 mutation status may be useful for risk assessment and breast carcinoma prevention in such families.

In addition, as noted above, most of the at-risk women in families without detectable BRCA1 or BRCA2 mutations were affected with breast carcinoma. These data support the general hypothesis that the number of breast malignancies per family may be not a sensitive means for identifying families with BRCA1 or BRCA2 mutations. These data also support the hypothesis that there are additional breast carcinoma susceptibility genes that remain to be identified, because the pattern of breast carcinoma in some of these families is consistent with the presence of a highly penetrant, autosomal-dominant susceptibility allele. Undetected BRCA1 and BRCA2 mutations may explain some of these families; however, linkage analysis and mutation-detection sensitivity estimates suggest that undetected mutations are unlikely to explain all of the families without mutations. Large genomic deletions, a known source of BRCA1 mutations, do escape detection by both DHPLC and direct sequencing. However, BRCA1 genomic rearrangements seem to account for only 10–12% of families with mutation-negative breast and ovarian carcinoma, with most of those rearrangements found in families with multiple primary breast and ovarian carcinomas.47

The results of this study reaffirm several phenotypic features that may be predictive for the presence of BRCA1 and BRCA2 germline mutations, such as ovarian carcinoma, male breast carcinoma, and 3 family generations with breast carcinoma diagnoses. This study also provides a combined analysis of both genes in a single cohort as well as information that may be useful in defining the prevalence, mutational spectrum, and penetrance of BRCA1/BRCA2 carcinoma susceptibility genes. All of this information should be considered when counseling patients about undergoing genetic testing and, in particular, when considering management strategies for women with documented mutations (for Sardinian patients with breast carcinoma, the high prevalence of the BRCA2-8765delAG variant justifies the screening of all women for such a founder mutation).

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors are grateful to the participating patients and families for their important contribution to this study, and they offer a special thanks to Dr. Assunta Criscuolo for data management.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES