Paget's disease of bone (PDB; OMIM 167250, 602080) is a chronic disease that typically results in enlarged and deformed bones in one or more regions of the skeleton.1, 2 Excessive bone breakdown and formation disrupt normal bone architecture and strength. As a result, bone pain, arthritis, noticeable deformities, and fractures can occur.
The etiology of PDB has remained largely unknown for several decades. Both morphologic and immunocytologic studies demonstrated the presence of paramyxovirus material in pagetic osteoclasts, suggesting that a latent viral infection may be involved in the pathogenesis of this disorder.2, 3 However, PDB also has a clear hereditary component. Familial clustering has been recognized to occur in PDB in 10% to 40% of cases, and epidemiologic studies have indicated that the relative risk of PDB in first-degree relatives of patients is about 7 to 10 times greater than in the general population.4–6 Genome-wide scan in families with PDB identified at least 7 potential susceptibility loci for the disease, even though some of these gene assignments turned out to be false-positives.7 In 2002, Laurin and colleagues identified a recurrent C→T transition at position +1215 leading to a proline-to-leucine substitution at codon 392 (P392L) on the SQSTM1 gene (within the 5q35 PDB3 locus) as a cause of PDB in about 50% and 20% of familial and sporadic French-Canadian patients, respectively.8 This gene encodes the p62/sequestosome 1 protein, which acts as a scaffold protein in the NFκB pathway as well as an intermediate protein in the proteosomal degradation of polyubiquitinated proteins. The same P392L mutation was identified subsequently in familial and sporadic PDB subjects from different countries.9–19 Currently, at least 20 further mutations in the SQSTM1 gene have been identified, all of which are clustered within or near the ubiquitin-associated (UBA) domain of the protein and lead to increased NFκB signaling and enhanced bone resorption. In some but not all patient samples, truncating mutations (where all or part of the UBA domain is deleted) were associated with a more severe phenotype than missense mutations.11, 16, 17, 19–21 Despite the fact that SQSTM1 mutations have been associated with a consistent number of familial PDB cases, incomplete penetrance has been described, and the prevalence of these mutations is low in sporadic PDB.20–22 Moreover, even in PDB families with SQSTM1 mutations, some affected relatives without the mutation were described, suggesting that additional factors (either genetic or exogenous) may be associated with disease expression.19 This is in keeping with results from experimental animal models of PDB.2, 23, 24
Marked geographic differences in the distribution of PDB also have been described, with a higher prevalence of the disease in populations of British descent.25 Moreover, increased-prevalence areas have been described in different countries. We recently characterized an area of increased prevalence of PDB in the region of Campania, in southern Italy. Patients from this region also showed increased severity of disease often associated with peculiar phenotypic characteristics (ie, giant cell tumor) and an increased number of familial cases.26–28 In this study we compared the clinical characteristics and prevalence and type of SQSTM1 mutations in a large sample of unrelated PDB patients from several Italian regions, including patients from the high-prevalence area of Campania. This sample also included three families with PDB associated with giant cell tumor. The large number of SQSTM1 mutations detected in our sample and the detailed clinical and anamnestic information collected from each patient have allowed us to better characterize genotype-phenotype correlation as well as to explore potential interactions between genetic and environmental factors.
Material and Methods
The participants in this study consisted of 608 unrelated and consecutive PDB patients from different Italian regions recruited at the Bone Disease Units of Turin, Siena, and Naples. These are the three main national centers for the diagnosis and treatment of PDB, located, respectively, in northern, central, and southern Italy (www.pagetitalia.com). General characteristics of recruited patients are reported in Table 1. All patients were born in Italy, and all but three patients were of Italian origin (as assessed from parental history). Diagnosis of PDB was based on biochemical evaluation, bone scintigraphy, and subsequent X-ray examination of areas of increased isotope uptake. For all subjects, a detailed medical history was obtained, including family history, place of birth, place of residence during childhood, occupation, age at diagnosis, skeletal extent, complications, age at onset of PDB symptoms, dietary habits, and animal contacts. The latter included either pet ownership or a lifestyle shared with animals in rural districts (ie, working or living on a farm). Specific questions were asked to assess the presence of animal contacts in different decades of life in each patient following a detailed questionnaire, as described previously.26 When available, clinical data were collected to evaluate the presence of PDB complications. In particular, the presence of neoplastic degeneration, cranial nerve disorders, hearing loss, hip or knee replacements, osteoarthritis, fractures, back pain, hypertensive disease, hyperparathyroidism, renal stones, or heart failure was recorded. The study was approved by local ethical committees, and all subjects had given informed consent to being included. All data were collected through common questionnaires shared by all participating centers. The cohort of PDB patients from Turin partially overlapped (54 of 186 patients) with a cohort examined in two previous studies on SQSTM1 mutations.13, 29
The patients having a previous report of at least one other family member affected with PDB were defined as familial cases. If the presence of relatives with suspected clinical features of PDB (ie, focal bone pain and/or bone deformity or deafness) was referred, these relatives were invited to undergo a specific diagnostic test for PDB. Patients with a negative history were classified as sporadic PDB.
First-degree relatives of all recruited patients also were invited to undergo biochemical evaluation of total alkaline phosphatase to uncover new familial PDB cases, which were confirmed by radiologic and bone-scan analyses. However, since we did not perform a detailed evaluation of all the first-degree relatives, we cannot exclude the possibility that sporadic patients may have had relatives with asymptomatic disease.
Pedigrees of the three families from Campania with PDB and giant cell tumor are shown in Fig. 1. All members from these families had polyostotic disease, with a high number of affected skeletal sites (7.6 ± 2.3, 5.0 ± 3.0, and 6.5 ± 2.1 in families 1, 2, and 3, respectively) and an early age of diagnosis (36.0 ± 9.5, 46.3 ± 1.5, and 32.5 ± 10.6 years in families 1, 2, and 3, respectively). Overall, five patients developed giant cell tumor. The mean ages at PDB diagnosis and giant cell tumor onset in these patients were 43.2 ± 9.3 and 58.8 ± 7.7 years, respectively, and the mean number of affected skeletal sites was 7.8 ± 3.5 (range 4 to 12). In two of these patients, multiple giant cell tumors were observed. After 12.0 ± 3.8 years from the diagnosis of giant cell tumor, four of these patients died owing to cardiovascular complications.
Genomic DNA was extracted from peripheral blood leukocytes using standard procedures. We conducted mutation screening of exons 7 and 8 of SQSTM1 and their intron-exon boundaries using PCR, followed by automated DNA sequencing. We performed PCR in reactions (25 µL) using Taq DNA polymerase (1 U; Fermentas, Glen Burnie, MD), 1X buffer, deoxynucleoside triphosphate (dNTP, 0.2 mM; Amersham, Uppsala, Sweden), primers (0.5 µM), and DNA (50 ng). PCR conditions were as follows: initial denaturation at 94°C for 3 minutes, followed by 35 cycles of 94°C for 30 seconds, 30 seconds at 62°C, and extension at 72°C for 45 seconds, and a final extension for 10 minutes at 72°C. Exons 7 and 8 of the SQSTM1 gene were amplified by using, respectively, two pairs of primers located in the flanking introns: 5'-CATGCGTGCTCCCCGACTGT-3'/5'-GCCCTGCAGTGGAGAACATC-3' for exon 7 and 5'-CTCTGGGCAGGCTCGGACAC-3'/5'-CTTGCACCCTAACCCCTGAT-3' for exon 8. Samples then were ExoSap-digested (Amersham) and sequenced using the Big Dye Terminator Ready Reaction Kit (Applied Biosystems, Foster City, CA). Sequencing reactions were performed on a 9700 Thermal Cycler (Applied Biosystems) for 25 cycles of 95°C for 10 seconds, 60°C for 5 seconds, and 60°C for 2 minutes. After the sequencing, each reaction was column-purified (Amersham) to remove excess dye terminators. Sequencing of the products was performed on the ABI Prism 3710 Genetic Analyser (Applied Biosystems). All the remaining exons and intron-exon boundaries of SQSTM1 also were screened in familial PDB patients who did not show mutations in exons 7 and 8.
To analyze the genetic background of mutated patients, we genotyped four single-nucleotide polymorphisms (SNPs) located in exon 6 (C916T, G976A) and the 3' untranslated region of SQSTM1 (C2503T, T2687G) as performed in previous studies.8, 22, 30, 31 The software program PHASE was used to reconstruct haplotypes (www.stat.washington.edu/stephens/).
In order to estimate the haplotype frequencies in the Italian population, as well as to differentiate the presence of SQSTM1 mutations from polymorphisms, we also analyzed DNA samples from 100 control subjects without any history of PDB or other skeletal disorders.
Continuous variables were compared by ANOVA. Comparisons between the groups were analyzed by chi-squared test or the Fisher exact test for categorical variables, whichever was appropriate. Analysis was performed using Statistica 5.1 (Statsoft, Tulsa, OK, USA) and SPSS (Release 6.1, SPSS, Chicago, IL, USA). All data are expressed as means ± SD. Adjustment for multiple comparisons was not performed.
General characteristics of patients
A family history of PDB in at least one relative was evidenced in 99 of the 608 recruited patients (16.3%). In keeping with our previous observations,27, 28 PDB subjects from Campania showed an earlier age at diagnosis (56.6 ± 11.3 versus 61.4 ± 12.1, p < .0001) and an increased clinical severity with respect to PDB patients from other Italian regions. In particular, an increased proportion of polyostotic cases (72.5% versus 50.4%, p < .0001) and an increased number of affected skeletal sites (3.1 ± 2.2 versus 2.2 ± 1.9, p < .0001) were observed in PDB patients from Campania than in patients from other regions. No correlation was observed between year of birth and number of affected sites in the overall sample of PDB patients. Conversely, the year of birth was negatively correlated with the age at diagnosis of PDB (r = −0.31, p < .0001). The latter association was highest in PDB patients living outside Campania (r = −0.65, p > .0001) and decreased in magnitude in PDB patients from Campania (r = −0.22, p < .005).
Mutation screening of the SQSTM1 gene in the overall sample
Genetic analysis was completed in 533 of the 608 patients. Eleven different mutations in the SQSTM1 gene were observed in 34 of 92 (36.9%) and 43 of 441 (9.7%) of familial and sporadic PDB patients, respectively (equivalent to 14.4% of the overall cohort) (Table 2). DNA analysis from 100 control subjects failed to detect the reported SQSTM1 mutations. A significantly higher prevalence of mutations was observed in polyostotic than monostotic patients (21.0% versus 5.7%, p < .0001).
Table 2. Prevalence of SQSTM1 Gene Mutations in Italy
Northern Italy (n = 132)
Central Italy (n = 169)
Southern Italy (n = 212)
Islands (n = 15)
With the inclusion of 1 patient with a G411S/P392L mutation, 5 patients without SQSTM1 mutations were excluded from analysis owing to the inability to assign their region of origin.
Two of these mutations, M401V (A1241G) and A427D (C1320A), were novel and have not been described previously. Moreover, we disclosed for both amino acids a high degree of conservation across species from fish to humans (data not shown), which argues in favor of an important role of these amino acids in the function of p62 protein. The other mutations (Y383X, P387L, P392L, E396X, M404V, G411S, D423X, G425E, and G425R) were described previously in other populations. One subject carried a double P392L and G411S mutation. He had a polyostotic form of disease (with three affected skeletal sites) diagnosed at 55 years of age.
The overall prevalence of SQSTM1 mutations was higher in younger than in older PDB patients. In fact, 64.9% of the mutations were observed in subjects with a birth age above the median (corresponding to 1937). The SQSTM1 mutation rate was 9.5% versus 19.2% in PDB patients below or above the median age (p < .01). A similar trend was observed when patients were grouped according to the decades or quartiles in relation to year of birth. Overall, 31 of 77 SQSTM1 mutations (40.26%) were observed in subjects in the upper quartile of age (year of birth after 1946).
After the analysis of exons 7 and 8, all the remaining exons and intron-exon boundaries of SQSTM1 were screened in the 58 familial PDB patients who did not show mutations in exons 7 and 8. No further mutations in SQSTM1 gene were found. In particular, all the three families with giant cell tumor complicating PDB were negative for SQSTM1 mutations.
As shown in Table 3, PDB subjects with SQSTM1 mutation showed an increased number of affected skeletal sites, an increased prevalence of polyostotic disease, and earlier age of onset than PDB patients without mutation. Conversely, no differences were observed in the occurrence of major complications of PDB between patients with or without mutation. A similar pattern also was observed in familial versus sporadic PDB patients. With respect to sporadic patients, familial patients were younger (63.5 ± 12.0 versus 69.3 ± 11.6 years, p < .005) and showed an earlier age at onset (54.6 ± 12.1 versus 60.7 ± 11.3 years, p < .0001), a higher number of affected skeletal sites (3.56 ± 2.7 versus 2.33 ± 1.8, p < .0001), and an increased proportion of polyostotic disease (73.8% versus 54.9%, p < .005). When familial patients with and without SQSTM1 mutation were compared, there were no differences in age at onset (53.9 ± 10.9 versus 54.9 ± 12.8 years, p = 0.7), and ther was a mild but not significant variation in the number of affected skeletal sites (4.07 ± 2.7 versus 3.23 ± 1.9, p = 0.08) or in the prevalence of polyostotic disease (67.2% versus 85.1%, p = .06). Conversely, PDB patients with SQSTM1 mutations considered to be sporadic cases showed an earlier age at onset (55.8 ± 11.5 versus 61.4 ± 11.2 years, p < .005), an increased number of affected skeletal sites (3.31 ± 2.3 versus 2.18 ± 1.8, p < .0005), and a higher prevalence of polyostotic disease (82.2% versus 50.9%, p = .0001) than sporadic PDB patients without SQSTM1 mutations.
Table 3. Clinical Characteristics of Patients With or Without SQSTM1 Gene Mutations
Familial patients (n)
Age (years ± SD)
Age at diagnosis (years ± SD)
Polyostotic patients (n)
Affected sites (n ± SD)
69.7 ± 21.5
60.6 ± 11.6
2.31 ± 1.9
67.8 ± 12.2
55.0 ± 11.2
3.60 ± 2.6
In order to better characterize genotype-phenotype correlation, we analyzed 98 first-degree relatives of subjects with SQSTM1 mutations, and we detected 20 additional mutations in 18 affected and 2 unaffected subjects. Interestingly, the mutation was not observed in 2 affected PDB relatives of two familial PDB patients with SQSTM1 mutations, indicating phenocopy. The first kindred was composed of two affected brothers from southern Italy, of whom only one had the M404V mutation. They had a similar polyostotic phenotype with three or four affected skeletal sites and a similar age at diagnosis (around 50 years). The second kindred included three affected patients from central Italy, of whom two had the P392L mutation. The two mutated patients, an 87-year-old woman and her 85-year-old brother, had polyostotic PDB with two and three affected sites, respectively. Their affected brother (82 years old) without a SQSTM1 mutation had monostotic PDB of the left pelvis.
Phenotype characteristics of subjects according to the type of SQSTM1 mutation are shown in Table 4. Overall, slight differences in the number of affected skeletal sites or age at disease onset were observed among patients with different mutations. A higher number of affected skeletal sites was observed in the two patients with A427D mutations (7.00 ± 2.8, range 5 to 9) and in the seven unrelated patients with Y383X mutations (4.84 ± 3.8, range 1 to 12). The latter mutation also was associated with the lowest age at diagnosis (mean 48.1 ± 9.3 years, range 40 to 60 years). Similar trends were observed when familial and sporadic patients were considered separately or when affected relatives with SQSTM1 mutations were included (bringing the overall number of mutated patients to 95). In this latter case, marked differences in clinical severity of disease also were observed, even within each single family with P392L, M404V, or E396X mutations. In fact, even in the case of a family with an E396X mutation, causing the truncation of most of the UBA domain, the number of affected skeletal sites in three PDB patients varied from two to seven, with an estimated onset of disease at between 38 and 64 years of age.
Table 4. Genotype-Phenotype Correlation in Patients with SQSTM1 Gene Mutations
Familial patients (n)
Age (years ± SD)
Age at diagnosis (years ± SD)
Polyostotic patients (n)
Affected sites (n ± SD)
63.5 ± 4.6
49.0 ± 8.8
4.28 ± 3.8
80.0 ± 14.0
64.3 ± 9.3
3.33 ± 2.3
69.5 ± 11.0
56.7 ± 11.8
3.56 ± 2.6
55.0 ± 13.6
50.5 ± 17.0
3.25 ± 2.5
68.2 ± 16.5
53.5 ± 11.5
3.07 ± 2.1
59.2 ± 7.2
49.7 ± 5.4
2.25 ± 0.9
59.5 ± 2.1
52.0 ± 5.6
3.50 ± 0.7
78.0 ± 7.0
62.0 ± 7.1
7.00 ± 2.8
To further explore possible genotype-phenotype correlations, we grouped patients according to the type or site of SQSTM1 mutation: truncating versus missense and outside versus inside the structured region of the UBA domain (amino acids 392 to 431). As shown in Table 5, an earlier age at diagnosis was observed in patients with truncating mutations than in those with missense mutations, whereas the number of affected skeletal sites and the frequency of polyostotic disease did not differ significantly. Mutations outside the UBA domain (Y383X and P387L, n = 10) did not differ significantly from mutations inside the UBA. Moreover, we observed a negative correlation between year of birth and severity of disease, expressed as number of affected sites, in patients with missense mutations (r = −0.22, p < .05) but not in patients with truncating mutations. The correlation between birth year and age at PDB diagnosis observed in the overall sample remained statistically significant independent of type (truncating or missense) or site (inside or outside UBA) of mutation.
Table 5. Genotype-Phenotype Correlations According to the Type (Truncating versus Missense) or Site (Outside versus Inside the Structured Region of the UBA Domain) of SQSTM1 Mutation
Regional distribution of SQSTM1 mutations and gene-environment interactions
The prevalence of SQSTM1 mutations was higher in southern (18.4%) than in central (11.8%) and northern (12.9%) Italy (Table 2). Moreover, an increased heterogeneity and a different distribution of mutations were observed in southern Italy (showing 9 of the 11 mutations) than in central and northern Italy (where only 4 and 5 of the reported mutations were observed, respectively). Interestingly, all seven patients with the Y383X mutation were from southern Italy and specifically from Campania. In contrast, the M404V mutation was more frequent in northern (4.6%) and central (3.0%) Italy than in southern Italy (0.5%).
Given the reported characteristics of PDB patients from Campania, we performed a subanalysis of patients from this region. Differences in terms of age at diagnosis and severity between familial and sporadic patients or between patients with or without SQSTM1 mutations were milder in PDB patients from Campania than in patients from other regions. Moreover, despite a higher prevalence of SQSTM1 mutations in the group of patients from Campania, a consistent number of analyzed PDB families from this region (26 of 35) did not have the mutation. Genotype-phenotype analysis in PDB patients from Campania did not evidence any significant difference in relation to the type or site of mutation. Moreover, the prevalence of SQSTM1 mutations did not differ significantly based on decades or quartiles of age in patients from Campania, whereas marked age-related differences in the prevalence rates of SQSTM1 mutations were observed in patients from the other regions (20.9% versus 8.4%, p < .001 in patients above or below the median age, respectively).
Overall, 342 of 533 (64.2%) PDB patients indicated animal contacts for at least 10 years before onset of the disease. These included pet ownership (mainly cats and dogs) and previous or current contact with animals such as pigs, rabbits, sheep, and cattle in rural districts. The prevalence of subjects with these contacts did not differ between patients with or without SQSTM1 mutations in the overall sample (58.4% versus 65.2%, p = 0.2) but became significant in patients living outside Campania (48.9% versus 64. 2%, p < .05). Conversely, patients from Campania had a high prevalence of animal contact (69.3% versus 62.1% in the other regions, p = .09) but without any difference between patients with or without SQSTM1 mutations. When familial patients from Campania were considered, animal contacts were observed in 90.1% of families without SQSTM1 mutations. Interestingly, a progressive age-related decrease in the prevalence of animal contact was observed based on the median age as well as the decade of age in the overall sample of patients. This decrease was not observed in the subgroup of patients from Campania but increased in magnitude in patients from other regions. In particular, in younger patients (with a birth year after 1950), 70.4% of patients from Campania reported animal contacts with respect to 48.9% of patients from other regions (p < .05).
Overall, PDB patients reporting animal contacts showed an increased number of affected sites (2.19 ± 1.9 versus 2.54 ± 2.0, p < .05) and a higher prevalence of familial disease (23.0% versus 14.1%, p < .05) than patients without animal contacts. A significant difference in the number of affected skeletal sites in relation to animal contacts also was seen in the subgroup of patients with SQSTM1 mutations (3.84 ± 2.5 versus 2.76 ± 2.2, p < .05). This difference also was seen when missense or truncating mutations were considered separately. All the preceding differences became milder and not significant when only pet ownership was considered instead of animal contacts.
SQSTM1 haplotypes in PDB patients and controls
We genotyped four SNPs in exon 6 and the 3' untranslated region of the SQSTM1 gene. The four selected SNPs were analyzed in all mutation carriers as well as in 100 control individuals and 100 PDB patients without SQSTM1 mutations. Genotype distribution for these SNPs followed a Hardy-Weinberg equilibrium. There was no significant difference in distribution of the genotypes between patients and controls for any of the SNPs studied.
Consistent with previous studies in different populations,8, 22, 30, 31 the H1 (916T-976A-2503C-2687T) and H2 (916C-976G-2503T-2687G) haplotypes accounted for the largest proportion of patients (94.3%) and controls (91.5%). The remaining patients were accounted for by six rare haplotypes with individual frequencies of between 0.2% and 3.6%. The presence of H1 and H2 haplotypes was observed in 75.0% (including two H1/H1 homozygous subjects) and 90.0% (including seven H2/H2 homozygous subjects) of patients with the P392L mutation, respectively, compared with 76% and 80% observed in control individuals. Since we did not perform allele-specific PCR, we could not unambiguously assign the mutation to one of the two haplotypes in the 26 H1/H2 heterozygous subjects, except than in three familial patients, in whom genetic analysis of affected family members was able to assign the P392L mutation to the H2 haplotype. An increased prevalence of the H2 haplotype also was observed in patients with M404V and Y383X mutations. In particular, the Y383X mutation was associated with the H2 haplotype in 100% of patients. In fact, only one heterozygous H1/H2 subject with Y383X was observed, and subsequent genetic analysis of affected family members (with the identification of one H2/H2 mutation carrier) demonstrated that the mutation is carried with the H2 allele. Conversely, 3 of 4 and 4 of 4 unrelated PDB patients with the G425E and E396X mutations, respectively, were negative for the H2 haplotype, suggesting that in this case the mutation is carried with a different haplotype, probably the H1, which was present in 100% of these patients (with a frequency of 87.5% and 75.0% in G425E and E396X mutation carriers, respectively).
Despite the significant progress that has been made in recent years, the etiology of PDB is not completely understood.2, 3, 24 Mutations in the SQSTM1 gene have been described worldwide in consistent proportion of patients with PDB from different ethnic groups, suggesting that functional differences in this gene are a direct cause of the disease, particularly in familial PDB. However, several clinical and experimental observations raised the hypothesis that other genes and/or environmental triggers are necessary to cause the disease, at least in some cases.2, 20, 21, 23, 32–34 In this study we report the results of the largest SQSTM1 mutation screening performed to date in consecutive familial and sporadic PDB patients.
In our sample, we identified 11 different SQSTM1 mutations in 14.4% of patients. This percentage is comparable with the results of mutational analysis studies performed in other countries, such as Great Britain (13.9%), France (12.8%), and Canada (19.8%).8, 9, 11, 17 Consistent with previous findings, the prevalence of mutations was highest (36.9%) in patients with a clear family history. Since we were not able to clinically exclude the presence of PDB in all first-degree relatives of recruited patients, we cannot exclude that a proportion of patients reporting no family history and thus classified as sporadic patients may have familial PDB. In particular, in sporadic patients with SQSTM1 mutations, we observed an increased severity of disease that is remarkably similar to that seen in familial patients. The latter observation may suggest that a consistent proportion of sporadic PDB patients with SQSTM1 mutations may indeed represent familial patients. This is in keeping with a recent report in a well-characterized sample of PDB patients from the United States showing absence of SQSTM1 mutations in sporadic patients.18 Of interest, we also observed a higher heterogeneity of SQSTM1 mutations in our sample of patients of Italian ancestry with respect to patients from other countries. In fact, we detected 11 different mutations. Together with the results from a recent additional study in an Italian population,29 15 different SQSTM1 mutations have been described in more than 800 Italian patients analyzed to date. An additional mutation (P364S) has been described in a single PDB family of Italian descent living in Australia.19 This heterogeneity is higher than observed in populations of British descent or in other European populations and might reflect the complex history of Italy as well as the several foreign invasions and dominations that occurred between sixth and nineteenth centuries. Moreover, in our study, a different distribution and a higher heterogeneity of mutations were particularly observed in southern Italy (showing 9 of the 11 mutations) than in central and northern Italy. While the M404V mutation was more frequent in northern and central Italy than in southern Italy, the Y383X mutation was observed in seven unrelated patients from Campania and was absent in patients from other regions. This mutation also was described in a previous Italian study in a family from Campania and in two sporadic patients of unreported origin29 but was not observed in more than 1000 PDB patients from other countries analyzed in previous studies. The mutation is one of the two known truncating SQSTM1 mutations located outside the UBA domain and seems to be associated with a severe phenotype in most patients. A similar phenotype has been described in one familial patient with the other mutation (K378X), and functional analysis confirmed that this mutation, leading to the complete elimination of the UBA domain, is associated with potentiated osteoclast formation and bone resorption in human primary cell cultures.16
Despite the severe phenotype and the earlier age at onset of disease, we did not find SQSTM1 mutations in the three kindreds with giant cell tumor. This complication represents a quite unusual clinical feature of PDB (described in fewer than 100 patients worldwide) and occurs mainly in patients with severe polyostotic disease, with a remarkably higher prevalence in patients from Campania.35–37 In fact, more than 50% of the patients described originated in or descended from ancestors who lived in this Italian region.27 Thus it can be speculated that a different gene is responsible for this particular variant of familial PDB, alone or in combination with an environmental trigger.38
Phenotype-genotype associations in PDB have not been investigated extensively, and the results from available reports are conflicting. This may reflect the limited number of patients with SQSTM1 mutations, except the P392L mutation, available in previous studies. Consistent with recent observations in two populations of British descent and in the French population, we confirmed that PDB patients with SQSTM1 mutations have a more extensive disease and an earlier age at diagnosis than patients without SQSTM1 mutations.11, 17, 19 This observation is in contrast to a previous report in a smaller sample of patients from Italy that did not show significant phenotypic differences in relation to the presence of SQSTM1 mutations, as well as between familial and sporadic patients.29 The prevalence of SQSTM1 mutations, however, was lower in that study (8.7%) than in our or in previous studies, most likely reflecting the reduced number of familial patients (12 of 357, equivalent to 3.4%). When we compared clinical characteristics of patients with different SQSTM1 mutations, we did not observe major differences in the number of affected skeletal sites or in the age at diagnosis, even though a trend for a more severe phenotype clearly was observed in most patients with the Y383X and A427D mutations. Since all patients with these mutations were from Campania, we cannot exclude the possibility that this finding is related to the overall increased clinical severity of disease observed in patients from this region. Moreover, a variable disease severity was observed among affected members of kindreds with the same SQSTM1 mutation for all mutations. These findings are in keeping with two previous detailed analyses of large PDB kindreds with SQSTM1 mutations of diverse racial or ethnic background showing high variability in intrafamilial expressivity of disease as well as incomplete penetrance.20, 21 Moreover, in one of these studies, offspring who inherited an SQSTM1 mutation from their parents were diagnosed with PDB later in life and had less extensive disease than their parents.21 Even though a slight reduction in clinical severity and a higher variation in the number of affected skeletal sites among family members were observed in missense with respect to truncating mutations, we can conclude that there are no major genotype-phenotype differences in relation to the type or site of SQSTM1 mutation. Together with the described cases of incomplete penetrance20–22 and the examples of phenocopy observed in this and other previous studies,19 these findings further reinforce the hypothesis that additional factors may be required to cause the disease (at least in a group of patients) as well as its skeletal extension in subjects with or without SQSTM1 mutations. In this context, the presence of somatically acquired SQSTM1 mutations in the pagetic bone cannot be excluded, even though this hypothesis remains controversial and probably restricted to a limited number of patients.39, 40
We and others have previously evidenced an association between PDB and animal-related factors, as well as a significantly higher prevalence of the disease in rural than in urban districts.26, 41–43 In this study, we also demonstrated that patients reporting persistent animal contacts for at least 10 years before the onset of disease have an increased number of affected skeletal sites and an increased prevalence of polyostotic disease. Interestingly, this association also was evidenced in patients with SQSTM1 mutations, suggesting an interaction between genetic and environmental factors. The observed differences, however, were small, and their clinical impact remains to be addressed in future prospective studies with larger numbers of patients with SQSTM1 mutations. In fact, even though adjustment for multiple comparisons generally is not required in this kind of study,44, 45 with a more conservative approach (ie, with Bonferroni's correction), some of these differences will no longer be significant.
Nevertheless, together with different previous epidemiologic reports, these data provide further evidence that infective agents may be involved in PDB, not only promoting the occurrence of the disease in some patients but also influencing its clinical severity. In this context, the progressive age-related decrease in the number of patients reporting animal contacts observed in this study is consistent with secular trends showing a decrease in both the prevalence and severity of PDB over time25, 46 and also might explain, at least in part, the parallel age-related increase in the prevalence of SQSTM1 mutations observed in our sample. In fact, it cannot be excluded that in patients without SQSTM1 mutations, the disease also may originate from contacts with an environmental factor (alone or in combination with additional genetic causes). Thus a reduced exposure to these animal-related environmental agents in recent years could have led to a relative increase in the frequency of PDB patients owing to SQSTM1 mutations. Consistently, this phenomenon was not observed in PDB patients from Campania, where a higher prevalence of animal contacts was observed even in recent years, reaching 90% in familial PDB patients without SQSTM1 mutations. To date, the nature of the possible environmental trigger remains unknown, but several reports suggested that viral infections of the paramyxovirus family may infect the osteoclast, inducing most of the cellular abnormalities of PDB.32, 33, 47–49 Indeed, in one of these studies, canine distemper virus not only induced NFκB activation and bone resorption but also markedly increased sequestosome 1/p62 gene expression in human osteoclast cells.33
Finally, results from our analysis evidenced an increased occurrence of the P392 mutation with the H2 than H1 haplotypes. This is consistent with the notion of a “founder effect” for this mutation, even if it was not possible to unambiguously assign the mutation to one of the two haplotypes in most of the heterozygous subjects, because we did not perform allele-specific screening. Moreover, we also observed two H1-H1 homozygous patients carrying the P392L mutation, suggesting that the same mutation has occurred independently at least twice, as observed previously in the French-Canadian population.8, 22 An increased prevalence of the H2 haplotype also was seen in patients with the M404V mutation, whereas the Y383X mutation was carried with the H2 haplotype in all patients. This also strongly supports the presence of a founder effect for this mutation. Conversely, most patients with G425E and E396X mutations were negative for H2 haplotype, suggesting that in this case the mutation is carried with a different haplotype, most likely H1.
In conclusion, while the clinical impact of most of the reported differences remains to be addressed, our findings further underline the complex etiology of PDB that cannot be explained solely by the presence of SQSTM1 mutations. It is likely that both genetic and environmental factors may cause PDB or most likely interact with each other to cause the disease and its variable phenotypes. Moreover, our results also indicate that the increased severity of PDB cases from Campania observed in this and other previous studies seems to be related to concomitant factors: (1) an increased heterogeneity of SQSTM1 mutations (with higher prevalence of truncating mutations such as Y383X), (2) an increased persistence of the environmental trigger (probably related to a lifestyle shared with animals), and (3) the presence of additional predisposition genes (including a gene causing familial PDB with giant cell tumor). Further genetic studies in this population, particularly in familial patients negative for SQSTM1 mutation, might be extremely useful for a real understanding of the complex etiology of this disorder.
LG and FG contributed equally to this work. All the authors state that they have no conflicts of interest.
The work was supported in part by grants from the Italian Association of Patients with Paget's Disease (AIP, www.pagetitalia.com) and the University of Siena (PAR 2006). The authors also thank Dr Vincenzo De Paola and Dr Annalisa Avanzati for technical assistance.