A Negative Search for a Paramyxoviral Etiology of Paget's Disease of Bone: Molecular, Immunological, and Ultrastructural Studies in U.K. Patients


  • Presented in part at the 18th meeting of the American Society for Bone and Mineral Research, Seattle, Washington, U.S.A., 1996, and at the 28th Meeting of the European Societies for Bone and Mineral Research, Maastricht, The Netherlands, 1999


Paget's disease of bone is a common bone disease characterized by increased and disorganized bone remodeling at focal sites throughout the skeleton. The etiology of the disease is unresolved. A persistent viral infection has long been suggested to cause the disease. Antigen and/or nucleic acid sequences of paramyxoviruses (in particular measles virus [MV], canine distemper virus [CDV], and respiratory syncytial virus [RSV]) have been reported in pagetic bone by a number of groups; however, others have been unable to confirm this and so far no virus has been isolated from patients. Here, we reexamined the question of viral involvement in Paget's disease in a study involving 53 patients with established disease recruited from seven centers throughout the United Kingdom. Thirty-seven patients showed clear signs of active disease by bone scan and/or histological assessment of the bone biopsy specimens and 12 of these had not received any therapy before samples were taken. Presence of paramyxovirus nucleic acid sequences was sought in bone biopsy specimens, bone marrow, or peripheral blood mononuclear cells using reverse-transcription polymerase chain reaction (RT-PCR) with a total of 18 primer sets (7 of which were nested), including 10 primer sets (including 3 nested sets) specifically for MV or CDV. For each patient at least one sample was tested with all primer sets by RT-PCR and no evidence for the presence of paramyxovirus RNA was found in any patient. In 6 patients, bone biopsy specimens with clear histological evidence of active disease tested negative for presence of measles and CDV using immunocytochemistry (ICC) and in situ hybridization (ISH). Intranuclear inclusion bodies, similar to those described by others previously, were seen in pagetic osteoclasts. The pagetic inclusions were straight, smooth tubular structures packed tightly in parallel bundles and differed from nuclear inclusions, known to represent MV nucleocapsids, in a patient with subacute sclerosing panencephalitis (SSPE) in which undulating, diffuse structures were found, arranged loosely in a nonparallel fashion. In the absence of amplification of viral sequences from tissues that contain frequent nuclear inclusions and given that identical inclusions are found in other bone diseases with a proven genetic, rather than environmental, etiology, it is doubtful whether the inclusions in pagetic osteoclasts indeed represent viral nucleocapsids. Our findings in this large group of patients recruited from throughout the United Kingdom do not support a role for paramyxovirus in the etiology of Paget's disease.


PAGET'S DISEASE is a metabolic bone disease characterized by focal areas of increased bone remodeling, resulting in a disorganized and weakened bone structure. The disease usually becomes manifest only after the age of 50 years and affects mainly the axial skeleton. Affected areas show increased numbers of osteoclasts and osteoblasts and the size of the osteoclasts and their number of nuclei are increased dramatically. Symptomatic patients suffer from bone pain, deformity, and pathological fractures. The disease is relatively common in parts of western Europe, the United States, and Australia where up to 3% of the population over the age of 55 years are affected,(1) but strong regional differences in incidence exist.(2,3) A particularly high prevalence of the disease occurs in the Lancashire region of the United Kingdom.(4) Despite intense investigation, the etiology of Paget's disease of bone is unresolved. For many years an environmental factor, in particular a slow viral infection with one of the paramyxoviruses, has been hypothesized as the cause of abnormalities seen in the bone cells, but recent work has focused on a genetic origin of the disease. (5–8) Consistent with either hypothesis, the prevalence of the disease has fallen in the United Kingdom and New Zealand over the past 25 years,(3,9) suggesting that the environmental factor causing the disease has changed in frequency over recent years, or, alternatively, pointing to dilution of the gene pool. A genetic component in the etiology of Paget's disease seems without doubt, but the question remains whether environmental factors, such as persistent presence of paramyxoviruses, are required to precipitate the disease. The electron microscopic detection of paramyxovirus nucleocapsid-like inclusion bodies in nuclei of pagetic osteoclasts by Rebel and coworkers(10,11) first prompted investigations into the presence of viral antigen or nucleic acid sequences. Using immunocytochemistry (ICC), in situ hybridization (ISH), or in situ reverse-transcription polymerase chain reaction (RT-PCR), a number of groups reported positive results in pagetic bone or bone marrow cells for either measles virus (MV), respiratory syncytial virus (RSV), or canine distemper virus (CDV). (12–16) However, no conclusive confirmation of viral presence was obtained using RT-PCR on bone-derived RNA. Although one laboratory consistently found MV in bone marrow and blood cells of patients, another found mainly CDV and three laboratories were unable to detect either virus. (17–21) Several arguments have been put forward to explain these apparent discrepancies, including the possibility that viral nucleic acid may be mutated and at low abundance and therefore difficult to detect by RT-PCR, at least from bone, or the possibility that regional differences may exist in the involvement of virus. Alternatively, positive detection of virus might have resulted from contamination by laboratory strains.

In this study we reexamined the question of viral involvement in Paget's disease using the highly sensitive technique of nested RT-PCR. We tested 37 bone biopsy specimens, 16 bone marrow samples, and 13 peripheral blood samples obtained from 53 Paget's patients recruited at seven centers in the United Kingdom using a total of 18 primer sets, 7 of which were nested. Primers were designed to detect various members of the paramyxovirus family, including MV, mumps, CDV, and RSV and included degenerate primers for conserved regions in the nucleocapsid and fusion protein genes to detect potentially mutated or novel paramyxoviruses. We also examined six of the bone biopsy specimens for the presence of MV and CDV using ICC and ISH. Finally, we compared the ultrastructural features of the inclusion bodies found in nuclei of pagetic osteoclasts in this group of patients with the inclusion bodies in brain cell nuclei from a patient with subacute sclerosing panencephalitis (SSPE), a persistent infection with MV, in which the inclusions are known to represent MV nucleocapsids.(22,23)



Transiliac crest bone biopsy specimens were obtained from 37 patients with radiological evidence for Paget's disease. In 15 patients qualitative histological examination of the biopsy specimen, carried out in parallel, showed active disease in the sampled tissue (increased numbers of osteoclasts and osteoblasts and presence of woven bone). In 17 patients, histological examination was not performed and in 5 patients, the histological data suggested inactive disease in the sampled tissue, despite a positive bone scan (two cases) or an elevated serum alkaline phosphatase (two cases). Eleven patients (all with demonstrated active disease in the sampled tissue) had not received any previous therapy, whereas others had received bisphosphonate therapy at some point before biopsy (in most cases the time since last treatment was more than 6 months). Bone marrow aspirates were obtained in 9 patients, all of which had moderately active disease as evidenced by histological examination of a previously taken (7 patients) or simultaneous (2 patients) bone biopsy. Three patients had not previously received any treatment, whereas the other 6 patients had received intravenous bisphosphonate at some point preceding the marrow aspirate with the time since last treatment ranging from 2 months to 1.5 years. In 7 patients bone marrow was obtained simultaneously at both an affected and a nonaffected site, giving a total of 16 samples for analysis. Peripheral blood was obtained from 13 patients, 4 of which had active disease in at least one skeletal site (as determined by bone scan). All but 1 patient had previously received bisphosphonate treatment with the time since last treatment ranging from 2 months to 3 years.

Bone marrow and peripheral blood samples were collected using EDTA as anticoagulant and the cells were separated on Ficoll gradients immediately after collection. The mononuclear cell fraction was used without further culture in the RT-PCR studies. Patients were recruited from seven centers in the United Kingdom as detailed in Table 1. As controls we examined bone biopsy specimens from 6 patients with renal bone disease and 1 patient with hyperparathyroid bone disease and osteophytes from 1 patient with osteoarthritis (all had high bone turnover as evidenced by increased numbers of osteoclasts and resorption surfaces in qualitative histology). Control blood samples were obtained from 7 patients with osteoporosis. All controls were recruited in Aberdeen, U.K. (Table 1). All patients gave informed consent for this study, which was approved by the local ethical committees from the participating centers.

Table Table 1.. Patient Information
original image

RNA extraction and RT-PCR

Bone biopsy specimens were snap frozen in liquid nitrogen and stored at −80°C until used. Total RNA was extracted from bone pulverized in liquid nitrogen or from blood or bone marrow mononuclear cell pellets, using RNAzol (Gibco, Paisley, U.K.). For mononuclear cell pellets glycogen was used as a carrier during RNA precipitation (Pharmacia, St. Albans, Herts, U.K.). Routinely, 2 μg of total RNA were reverse transcribed using Superscript reverse transcriptase (Gibco) and a random hexamer primer (2 μg per reaction; Boehringer, Lewes, East Sussex, U.K.) using the manufacturer's instructions. The final volume after RT was made up to 100 μl with H2O and complementary DNA (cDNA) stored at −20°C. PCRs were carried out in a volume of 25 μl, using 2.5 μl cDNA (approximately 50 ng) and 1 μM (final concentration) primer mix (2 μM for degenerate primers), 0.6 U of Taq polymerase (Advanced Biotechnologies, Epsom, Surrey, U.K.), and 1.5 mM MgCl2. The following cycling protocol was used: 94°C for 50 s, annealing temperature (Table 2) for 60 s, and 72°C for 90 s for 35 cycles. In the first cycle, the 94°C step was extended to 4 minutes and in the last cycle, the 72°C step was extended to 10 minutes. Primer sequences, GenBank accession numbers used to design the primers, product sizes obtained, and optimal annealing temperatures are given in Table 2. As control for extraction of nuclear RNA we used intronic primers for type I collagen as previously described.(24) PCRs were carried out on cDNA obtained in the presence and absence of reverse transcriptase in the RT step to control for a possible contribution from contaminating DNA. As controls for adequate RNA extraction from bone cells and successful RT we used primers for β-actin and β2-microglobulin (β2M) to detect cellular RNA, primers for tartrate-resistant acid phosphatase (TRAP V), and calcitonin receptor (CTR) to detect osteoclast-derived RNA and primers for osteocalcin (bone gla protein [BGP]) for osteoblast-derived RNA (Table 2). Two bone samples were excluded from the analysis because of consistent poor amplification of control products. Nested PCRs were carried out as described previously using 1 μl undiluted and 2.5 μl diluted (1:100 and 1:1000 in H2O) first-round PCR products as templates. PCR products were electrophoresed on a 1.5% agarose gel, stained with ethidium bromide, and photographed with a Herolab UV gel documentation system. The RNA extraction, RT, and assembly of the first-round PCR were performed in a laboratory (RNA lab) physically separated from the one in which the PCR cyclers were housed and the PCR products were analyzed (PCR lab). Separate sets of pipettes were employed and barrier tips were used for cDNA templates. No PCR products were allowed to enter the RNA lab. Reagents for the nested PCRs were assembled in the RNA lab and template was added later in the PCR lab (all using dedicated pipettes and barrier tips).

Table Table 2.. Primer Used in This Study
original image

Cloning and sequencing of PCR products

PCR products were analyzed on a 1.5% agarose gel. Where single clean bands were observed on the gel, a second round of amplification was carried out using a 100-fold dilution of the original PCR product as template. The resulting product was cleaned using the Qiaquick PCR purification kit (Qiagen, Crawley, West Sussex, U.K.) and sequenced directly using the ABI PRISM cycle sequencing kit (Perkin Elmer, Warrington, U.K.) and the specific product primers. For all other PCR products, bands of interest were excised from gels and eluted using the Qiagen gel elution kit (Qiagen). Gel-purified products were subjected to a second round of amplification using the original primers, and the resulting PCR products were cloned into PCRII (Invitrogen, Leek, The Netherlands) using the manufacturer's instructions. Positive colonies were identified by PCR using the original product-specific primers, and plasmid DNAs were prepared and sequenced as described previously using primers for SP6 and T7 bacteriophage promoter sites within the SPII vector. Sequences were compared with known viral sequences in the GenBank database(25) using the BLAST program.(26)

Validation of primers and sensitivity of nested PCR on RSV

All virus primers were validated by running PCRs on virus-containing samples. These reactions were carried out by one of us (R.P.H.) not involved with the patient samples and in a separate laboratory. No virus-infected material or viral PCR products were ever allowed in the laboratories in which the patient samples were analyzed. The control templates consisted of measles, mumps, and rubella (MMR) vaccine (Pasteur Mérieux, Maidenhead, U.K.; used for measles, mumps, and NC and FP degenerate primers), CDV vaccine (NOBI-VAC Puppy DP; Intervet U.K., Cambridge, U.K.; used for all CDV-specific primers and the MCFP primers which detect MV and CDV FP), and RSV-infected Hep-2 cells (for RSV primers). The sensitivity of the nested PCR compared with PCR using outer primers only was examined by template dilution for all nested primer sets. In the case of RSV, the detection limit of the nested PCR also was determined by serial dilution of known numbers of RSV (A2 strain)-infected Hep-2 cells, into uninfected Hep-2 cells. Ratios of infected-to-uninfected cells were calculated by immunofluorescence using a polyclonal antibody to RSV (Glaxo Wellcome, Beckenham, U.K.). Total RNA was extracted from each sample (all samples contained 105 Hep-2 cells) and reverse transcribed using a random primer and, for comparison, an oligo-dT primer. Nested PCRs were carried out using 1:100 dilutions of the products obtained with outer primers as template in the nested reaction.

Histology and ultrastructure

For light microscopy part of the bone biopsy specimen was fixed in 4% paraformaldehyde in phosphate buffer (0.1 M, pH 7.4) for 24 h at room temperature, followed by demineralization in the same fixative containing 2.5% EDTA (pH 7.2) for several weeks (until tissue had demineralized fully). Tissues were then dehydrated in ethanol and embedded in paraffin and sections (5 μm) were picked up on RNAse-free Superfrost Plus, slides (Merck, Poole, Dorset, U.K.). Great care was taken to change the blade after each block and all handling of sections was done under RNAse-free conditions.

For electron microscopy, part of the bone biopsy specimen was fixed in 2.5% glutaraldehyde in phosphate buffer (0.1 M, pH 7.4) for a minimum of 24 h, demineralized in the same fixative with 2.5% EDTA for approximately 48 h, washed in buffer, postfixed in osmium tetroxide, dehydrated in ethanol, and embedded in Epon (Taab, Aldermaston, U.K.). Ultrathin sections were stained using uranyl acetate and lead citrate. Several ultrathin sections of Epon-embedded brain tissue from a patient with SSPE(23) (gift of Dr. J. Martin, Antwerp, Belgium) were cut and stained as described previously. All ultrathin sections were examined with a Philips EM300 microscope and photographs were taken at the same magnification for the pagetic and SSPE samples.


To detect MV, dewaxed sections were stained with mouse monoclonal antibodies to MV nucleocapsid protein (clone 99-21, used at 1/2000 dilution; Harlan, Sera-Lab, Loughborough, U.K.). Bound antibody was detected using fluorescein isothiocyanate (FITC)-labeled secondary antibodies (DAKO, Ely, U.K.) and sections were examined using a Leica confocal laser scanning microscope. Dewaxed sections also were subjected to ISH with a digoxigenin (DIG)-labeled RNA probe to the nucleocapsid sequence of the MV genomic strand using previously described methods.(27,28) Hybridization was detected using an alkaline phosphatase-labeled anti-DIG antibody and the label developed using vector red as substrate (kit SK-S100; Vector Laboratories, Loughborough, U.K.). In one instance in which both ICC and ISH gave positive results on the same sections, confirmation of viral presence was sought using RT-PCR on RNA extracted from consecutive sections,(29) using primers for the phosphoprotein of MV (forward 5′ATG TTT ATG ATC ACA GCG GT3′ and reverse 5′ATT GGG TTG CAC CAC TTG TC3′) and as controls, primers for β-actin as described.(30) CDV was detected in dewaxed sections using a monoclonal antibody (D110) to CDV that detects all known virulent and laboratory strains.(31,32) Sections were treated for 2 h at 37°C with primary antibody, followed by goat-anti-mouse antibody and a peroxidase-antiperoxidase (PAP) complex. The label was developed using hydrogen peroxide and diaminobenzidine (DAB) as substrate. A rabbit polyclonal antibody to CDV also was used (antibody raised by one of us [S.L.C.] by immunization with CDV) and visualized with an FITC-labeled secondary antibody and confocal microscopy as described previously. ISH for CDV was performed as described previously.(32,33) A DIG-labeled RNA probe recognizing the nucleocapsid protein messenger RNA (mRNA) of the virulent CDV was used. After intensive washing to remove excess nonbound probe, slides were incubated with an anti-DIG antibody conjugated with alkaline phosphatase and the enzyme developed using nitroblue tetrazolium (NBT) and X-phosphate (Boehringer Mannheim, Lewes, East Sussex, U.K.). In all cases positive control tissues were included (brain tissue from a CDV-infected dog and brain tissue from a patient with SSPE) and negative controls consisted of staining/hybridization with omission of the primary antibody or probe, and, for ISH, hybridization of slides treated with RNase to remove the target sequence. Sections from patients and controls were processed simultaneously in a blinded fashion and where results were unclear, staining was repeated on freshly cut sections from the same tissue block and where possible on sections of other blocks from the same patient.


Primer validation and sensitivity of nested PCR

The optimum annealing temperatures for the sequence-specific primers for MV, mumps, RSV, and CDV and the degenerate nested primer for the nucleocapsid protein (NC2 and −3) were determined by amplification of cDNA prepared from MMR vaccine (MV, MPS, NC, and FP degenerate primers), RSV-infected Hep-2 cells (RSV), and distemper vaccine (CDV) and are given in Table 2. All products obtained were sequenced and results confirmed amplification of the expected nucleic acid sequences (not shown). Optimal conditions for TRAP and CTR primers, used as control for adequate extraction of osteoclast RNA, were established using osteoclastoma cDNA as template and identity of the products was confirmed by DNA sequencing (not shown). Primers for BGP and β-actin have been described and validated previously.(18,34) The relative sensitivity of the nested primer sets was determined by comparing the cut-off point for amplification from serial dilutions of template with the outer primers with the cut-off point for amplification with the nested primers (on 1:100 dilutions of template obtained in the first round of amplification). The relative sensitivity for nested primers was 10-100 times higher than for amplification with outer primers only (Fig. 1A). Most nested primer sets were more efficient in amplification from control templates than the outer primer sets (presumably because they amplify shorter fragments) and therefore both outer primers and nested primers were run in the first-round screening of patient samples. Confirmation of integrity of RNA and amplification of bone-derived cDNA were obtained in all samples before the virus screening was performed (Fig. 1C). The detection limit of the nested PCR for RSV was calculated from amplification of serial dilutions of RSV-infected cells and shown to be one RSV-infected cell in a background of 105 uninfected cells (Fig. 2).

Figure FIG. 1.

(A) Comparative sensitivity of nested primers versus non-nested primers. The example shown is for MPS primers (MPS1+4, followed by MPS2+3). Template dilutions were made in H2O. Arrowheads indicate presence of the correct product (confirmed by sequencing). M, size markers (phiX174/Hae III digest); pd, primer dimer. (B) Sensitivity of all nested primer sets compared with the non-nested primers. FP primers (semi-nested) produced only faint bands and their relative sensitivity was not determined. (C) Control products after 35 cycles of amplification in three individual bone samples. CTR = 212 base pairs (bp), TRAP = 298 bp, BGP = 408 bp and β2M = 323 bp. M, size markers.

Figure FIG. 2.

Determination of detection limit of the nested PCR for RSV. All samples contained cDNA from 105 Hep-2 cells infected with varying amounts of RSV. The number of RSV-infected cells was counted after immunostaining in three cultures inoculated with reducing amounts of infected cells (A, lanes 1-3). Further dilutions of infected cells were made in noninfected Hep-2 cell suspensions (A, lanes 4-8). There was no difference in sensitivity using random primed versus oligo dT primed cDNA (compare both panels in A). Non-nested PCR detects RSV down to 10 infected cells in a background of 105 uninfected cells (A, lane 4). (B) Nested PCR is 100-fold more sensitive, with a clearly positive signal in lane 6. Note how template overload in the nested PCR leads to reduced amount of product (B, lanes 1 and 2). (C) Roughly, equal amounts of β2M products are obtained from the templates used in the first-round PCR, indicating that equal amounts of Hep-2 cells were present in each lane. M, size markers; pd, primer dimer. (D) Shows immunostaining of RSV-infected Hep-2 cells (arrowhead); negative cells are indicated by an asterisk.

RT-PCR on patient and control samples

In all bone and bone marrow samples β2M amplification product was easily detectable after 35 PCR cycles. A product for BGP also was obtained in all bone samples, indicating that osteoblast-derived RNA was present, but in marrow samples, where osteoblast numbers would be expected to be low, the BGP signal was weak (Table 3; Fig. 3B). TRAP mRNA was amplified in 34/36 bone samples and 15/16 marrow samples, indicating presence of osteoclast-derived RNA. A product for CTR was obtained in 24/36 bone samples and 2/16 marrow samples. For CTR two rounds of amplification were regularly required to obtain product visible on agarose gels (for examples of controls, Figs. 1C and 3). Amplification with intronic primers for type I collagen showed product visible on agarose gels in 5 out of 10 samples tested (not shown). No product was observed when the reverse transcriptase was omitted from the RT reaction, excluding amplification from contaminating DNA. All bone biopsy specimens and bone marrow samples were analyzed with the full set of primers (including the nested primers) in a first round of amplification, followed by PCR on 1:100 and 1:1000 dilution of the reaction products with the appropriate nested primers (Fig. 3A). In none of the reactions were products of the expected size observed. A number of products of unexpected size were observed regularly with CDV1+493, NC2+3 and RSV2+3 (Fig. 3B) in the first round of amplification; however, no amplification was obtained from CDV1+493 product with the nested primers CDV2+3, indicating that the product obtained is not CDV. Furthermore, these products of unexpected size also were seen in the control samples (not shown) and on sequencing revealed no homologies with virus or any other sequences in GenBank and are likely to be products of nonspecific amplification (Table 3). Multiple bands were obtained with two of the specific primers for CDV (CDV2+493 and CDV488+489) and the semi-nested set for FP from bone and marrow templates, but no bands of the expected and previously reported size(12) were seen (not shown).

Figure FIG. 3.

(A) Example of a nested PCR on a bone sample. Note absence of any amplification with primers to viral sequences, even in the nested reaction, whereas control transcripts were detected easily, including CTR, which in this sample required a second round of 35 cycles of amplification. M, size markers; pd, primer dimer. (B) Examples of products obtained from two bone samples and one bone marrow sample in first-round PCR with 35 cycles. Some products of unexpected size are indicated, with the expected product size given in brackets. These products were sequenced and found to have no viral homology. Note how in bone marrow no clear product for BGP and CTR is obtained, consistent with absence or very low presence of mature bone cells. M, size markers. (C) No amplification of MV or CDV from peripheral blood mononuclear cells. Shown are four samples (designated 1-4) tested with each primer pair. The template was 500 ng of cDNA for each PCR and 45 cycles of amplification were used. As a control for each reaction 50 ng of cDNA was amplified in parallel using primers for β-actin. M, size markers.

Table Table 3.. RT-PCR Results
original image

The peripheral blood mononuclear cells were analyzed with the nested MV and CDV primers, and with MV5+6 and MV7+8 (2 rounds of 35 cycles). Products of the expected size were not observed in any of the reactions (Table 3).

Because these negative findings in RT-PCR might have been caused by very low abundance of target sequence in our samples and previous positive reports of MV detection by RT-PCR used higher concentrations of cDNA template in the PCR,(35) we repeated the analysis for MV on six bone, three bone marrow, and all peripheral blood cDNAs and we repeated the analysis for CDV on six bone samples. We used 250-500 ng of cDNA per PCR (i.e., 5-10 times as much as previously) and amplified this using MV5+6, MV7+8, and CDV 1+493 for 45 cycles or 2 rounds of 35 cycles. In these reactions, again no specific bands for MV or CDV were seen and often smearing was observed, indicating template overload (Fig. 3C).


The results of screening for MV and CDV by ICC and ISH are summarized in Table 4 and illustrated in Fig. 4. Consistent positive signals were not obtained using both techniques in any of the pagetic samples. One patient showed faint staining for MV in a number of bone marrow cells by ICC but not by ISH. One control was positive for MV by both ISH and ICC (very faint staining) in a small number of bone marrow cells ( Figs. 4D and 4E) but not by RT-PCR on RNA extracted from the sections and not by RT-PCR on bone biopsy-derived RNA. In some sections extensive staining of the bone matrix was observed with MV antibody (Fig. 4F). In each batch of tissues tested, positive control tissue (brain from a patient with SSPE and brain from dog with distemper) was included (Figs. 4C, 4G, and 4H). We also included a block of CDV-infected dog brain, which was fixed, demineralized, and processed for the same length of time as the test samples. All controls were consistently positive but the CDV ICC staining was lighter after prolonged demineralization. All staining was performed in duplicate and the CDV ICC also was performed after extended proteinase K digestion to aid in the unmasking of target sequences. Only in 1 patient was a positive signal observed by ISH for CDV in marrow cells (Fig. 4I), but not in osteoclasts or any other bone cell types. However, the staining was seen only in one block out of three from the same patient, and a second series of sections from the same tissue block was negative. All other samples were consistently negative (Fig. 4J).

Figure FIG. 4.

ICC and ISH on demineralized, paraffin embedded bone and control tissues. Bars in A and J represent 100 μm, bars in B-I represent 10 μm. (A) Light microscopic view of a typical pagetic bone biopsy specimen. Large, actively resorbing osteoclasts are seen (arrows) and areas of bone formation (arrowhead). Stained with hematoxylin and eosin. (B) Detail of typical pagetic osteoclasts. Note excessive numbers of nuclei. Stained with hematoxylin and eosin. (C) Neuron stained positive (FITC label, green) with MV antibody in a section of brain from a patient with SSPE. Nuclei are stained with propidium iodide (orange). (D) Light staining for MV by ICC (FITC label, green) in a number of cells in the bone marrow of a control sample from a patient with hyperparathyroid bone disease (control 4ROD). Arrows indicate positive cells. (E) Light staining for MV by ISH (red) in a number of cells in the same control sample as shown in D (control 4ROD). No signal was observed when hybridization with an RNA probe to the nucleocapsid gene was performed after treatment of the section with RNase. However, no confirmation of MV presence was obtained by RT-PCR on RNA extracted from further sections from the same block (not shown). (F) Strong staining by MV antibody of the bone matrix (FITC, green) in this patient sample (PD3). No staining is seen in any bone cell type. No cellular staining was observed with MV antibody in any of the Paget's samples tested by ICC or by ISH (not shown). Nuclei are stained with propidium iodide (orange). (G) Positive staining for CDV by ISH in distemper dog brain. The probe hybridizes with NP protein mRNA and is detected using alkaline phosphatase and NBT X-phosphate (purple), bright field, and no counter stain. (H) Positive staining for CDV in distemper dog brain by ICC. The antibody is detected using the PAP technique and developed using DAB (brown), bright field, no counterstain. (I) Positive staining for CDV by ISH in bone marrow cells in a discrete area in the only Paget's patient in whom positive cells were seen (arrows). No signal was observed in any bone cell type. A repeat on a fresh section from the same block was negative. Developed using NBT X-phosphate (purple), bright field image, and no counterstain. (J) Absence of staining for CDV by ISH in pagetic osteoclasts (arrows). No staining was observed in any bone cell type in Paget's patients or controls. Developed using NBT X-phosphate (purple), bright field image, and no counterstain.

Table Table 4.. ICC/ISH Results
original image

Electron microscopy

Viruslike inclusion bodies, as described previously in Paget's disease, were recognized easily in osteoclasts in four samples (Figs. 5A and 5B). In 10 other samples inclusions were not immediately obvious, but no exhaustive search was performed. Inclusions were found predominantly in the nuclei, although in one case both nuclear and cytoplasmic filaments were seen. Nuclear inclusions were not observed in other cell types or in osteoclast nuclei of a patient with hyperparathyroid bone disease, where a large number of osteoclasts were examined. The nuclear inclusions in pagetic osteoclasts were all highly similar and consisted of straight tubular structures with a diameter of around 15 nm. In some cells the inclusions occurred in nuclei that showed signs of degeneration (widening of space between nuclear membranes), although the rest of the cell seemed intact. Analysis of the inclusions found in nuclei in brain cells in a patient with SSPE showed that these structures were similar in size but had a more diffuse appearance and they clearly showed transverse striations. Their organization was different from the inclusions in Paget's disease. Although in Paget's disease the tubular structures were packed tightly into parallel bundles, in SSPE undulating structures were seen, organized loosely in a random fashion. (Figs. 5C–5F).

Figure FIG. 5.

TEM from typical inclusions in pagetic nuclei, compared with inclusions in brain cells from a patient with SSPE. (A) Low magnification of a multinucleated osteoclast with inclusions identified in one nucleus (arrow). Bar represents 20 μm. (B) Detail of nuclear inclusion from A. Bar represents 0.1 μm. (C and E) An overview and a high magnification of inclusions in a nucleus of a pagetic osteoclast, whereas D and F show inclusions in a brain cell of a patient with SSPE. Note the difference in organization of the structures and the “stiffness” of the pagetic inclusions compared with the undulating appearance of the MV nucleocapsids in SSPE. The micrograph pairs in C and D (bar represents 0.5 μm) and E and F (bar represents 0.1 μm) were taken with the same microscope, at the same magnification and printed with the same settings. All sections were counterstained with uranyl acetate and lead citrate.


In this study we found no evidence for the presence of paramyxovirus in 53 patients with Paget's disease using RT-PCR, including nested RT-PCR on either bone biopsy specimens, bone marrow aspirates, or peripheral blood mononuclear cells. Our current results confirm earlier observations by Ralston et al.,(18) Birch et al.,(17) and Ooi et al.(21) using a less sensitive non-nested PCR technique. Our study was designed to address most of the criticisms raised against these earlier studies. First, to overcome the potential problem of failure to detect very low copy numbers of a target sequence, we used nested RT-PCR, which showed an increased sensitivity over PCR with non-nested primers of 10- to 100-fold. Second, because it has been suggested that viral RNA may be more difficult to extract and amplify from bone tissue than from bone marrow or peripheral blood mononuclear cells, all types of samples were included. We did not find evidence for presence of specific inhibitors of RT-PCR in bone, because in control reactions several target sequences from bone-derived RNA, including less abundant, osteoclast-derived sequences such as CTR, were amplified consistently. Also, in previous studies from our group we have not encountered specific difficulties in detecting a variety of mRNA species from bone.(34) Difficulties in amplification of viral sequences caused by excessive secondary structure of transcripts and inefficient isolation of genomic sequences also have been suggested as a possible reason for negative findings in studies using RT-PCR.(12) However, our control reactions with virus-infected cells showed easy amplification of viral sequences (presumably from transcripts and genomic sequences). We did not expect to find paramyxovirus nucleic acid sequences in the nucleus, because amplification and replication of these viruses are known to occur exclusively in the cytoplasm. However, amplification and detection of unspliced RNA from the samples showed that nucleus-derived sequences, if present, would have been detectable in half of the cases. Therefore, we are satisfied that the techniques for RNA extraction and cDNA amplification used in this study were working adequately. Third, because mutations in the regions chosen for amplification of the various paramyxovirus family members could preclude successful amplification by RT-PCR, we used a much larger number of primers than in any study previously reported, including degenerate primers to conserved sequences between various paramyxovirus family members and covering two viral genes, the nucleocapsid gene and the fusion protein gene. We also included primers used previously by Gordon et al.(12) for detection of CDV and the primers used by Reddy et al.(14) for detection of MV. In addition, the nested primer sets were designed to amplify a region of the viral genome close to the sequences previously found in Paget's disease. Fourth, because it might be argued that because of sampling of only a small biopsy specimen, virus was absent in the material tested, despite being present in the affected lesions, we included a large number of patients (n = 53). In previous positive studies, using ISH, ICC, RT-PCR, and in situ RT-PCR, a high (70-100%) incidence of viral presence has been detected,(13, 14, 16, 36) which makes it unlikely that sampling bias could have caused our negative results. This also is unlikely because where biopsy material was examined by histology or transmission electron microscopy (TEM), clear evidence of active Paget's disease was seen in 15 cases, and in 4 patients, osteoclasts with nuclear inclusion bodies were detected. Regional variation in viral involvement could be an important factor; therefore, bone biopsy specimens were obtained from seven centers in the United Kingdom, and blood and bone marrow samples were obtained from two centers. It might be argued that approximately half of the patients examined in this study had at some point before obtaining bone, blood, or marrow samples been treated with bisphosphonates and that this might have affected the numbers of virally infected osteoclasts or osteoclast precursors. However, even if we eliminated all treated patients, we still had a large group of untreated patients that tested negative for virus by a variety of techniques. Also, in most treated patients there still was clear evidence of active disease on histological examination (with large numbers of osteoclasts) by bone scan or by clinical assessment. It is well established that bisphosphonate therapy does not provide a cure for Paget's disease with most patients experiencing relapse 18-24 months after treatment. This suggests that the causative agent must persist, even in treated patients, and should have been detectable in several of these. Finally, although in our study all material for TEM was obtained from untreated patients and therefore no conclusions could be drawn on persistence of nuclear inclusions (the putative viral nucleocapsids), several other studies have reported that nuclear inclusions are present in all patients, regardless of previous treatment with bisphosphonates.(10, 37, 38)

The initial interest in a viral etiology for Paget's disease started when Rebel and coworkers(39) and Mills and Singer(37) described the presence and ultrastructure of inclusion bodies in osteoclast nuclei. Nuclear inclusions have been found consistently in all patients with Paget's disease and their occurrence seems confined to osteoclasts. The ultrastructural features of the inclusions have been described to be less undulating and “stiffer” than those seen in SSPE,(37) and, on the basis of their dimensions, to bear more resemblance to pneumovirus (RSV) than to morbillivirus (MV).(40) Here, we also found that the organization of the pagetic inclusions in parallel arrays was markedly different from the more random way in which MV nucleocapsids are organized in SSPE, a disease proven to be caused by a persistent infection with MV.(41) Inclusions, highly similar to those seen in Paget's disease, have been described in osteoclast nuclei in a number of other bone diseases associated with osteoclast malfunction, such as osteopetrosis,(42,43) pycnodysostosis,(44) and osteoclastoma,(45) the first two of which have established genetic rather than environmental origins.(46,47) Although a viral infection could be a secondary phenomenon rather than the primary cause of the diseases described previously, the evidence for a true viral nature of the nuclear inclusion is unconvincing so far. Although in some cases there is correlation of presence of inclusions and immunocytochemical detection of virus, some patients show viral antigens but no inclusions.(42) A persistent infection with paramyxovirus requires viral transcription and replication; however, the inclusions have been observed most frequently in the nucleus, rather than in the cytoplasm where all these events take place. In SSPE, the occurrence of nuclear MV nucleocapsids has been suggested to occur only because of defective nucleocapsid genes, which lead to accumulation of “empty” nucleocapsids in a site not normally occupied by virus.(41) Also, despite the frequent occurrence of inclusions in pagetic osteoclasts, viral budding has been suggested only in one unconfirmed report.(48) In the absence of definitive proof that inclusion bodies in Paget's disease indeed represent viral nucleocapsids and because ultrastructurally identical inclusions have been observed in osteoclasts in bone diseases known to have a genetic origin, it seems more likely that they represent some other feature of abnormal osteoclasts. The common feature between all disorders mentioned earlier is that the osteoclasts have increased numbers of nuclei, suggesting increased or continuing fusion of osteoclast precursors and/or reduced osteoclast death. Rather than representing a foreign body, the nuclear inclusions might equally be considered to represent ultrastructural features of nuclear matrix reorganization associated with such cellular events in multinucleated cells. Interestingly, the inclusions described in Paget's disease also bear resemblance to those seen in inclusion body myositis (IBM), a disease with an as yet unknown etiology. In IBM, the inclusions are present in multinuclear cells (the skeletal muscle cell), although their occurrence is predominantly cytoplasmic.(49) As in Paget's disease, the finding of the inclusions prompted a search for paramyxovirus. Initially, MPS virus was implicated based on ICC and ISH results; however, subsequent studies failed to confirm a causative role for MPS(50,51) and no virus has been isolated to date.(52)

To many, it may seem difficult to understand that with the currently available sensitive molecular tools, controversies over viral presence as causative agent in diseases cannot be resolved. However, recent debates on the etiology of multiple sclerosis, IBM, type II diabetes, Crohn's disease, and multiple myeloma (53–55) have illustrated that the exquisite sensitivity of molecular methods such as RT-PCR and in situ RT-PCR, particularly when used at their limits of sensitivity,(56) can easily lead to conflicting results. Contamination of test samples with laboratory strains of virus or plasmids carrying viral nucleic acid inserts or previously amplified viral products from control samples can occur easily, even where stringent controls are in place.(57) Also, immunological reagents used for virus detection may have previously unknown cross-reactivity with tissue antigens or with nonviral antigens, which could be expressed differently in the diseased tissue than in the control. In the case of Paget's disease it is striking that the two groups consistently reporting virus have only found the virus they have worked on most in their laboratories(12,14) and there is no convincing evidence for the detection of virulent, rather than laboratory strains, in the positive cases. We appreciate that a negative finding, as in this study, may be caused by insufficient sensitivity or another technical problem and we acknowledge that we did not establish the detection limit of our RT-PCR technique for all viruses tested. We also did not test our samples by in situ RT-PCR, which may be more sensitive than direct ISH as performed here. However, we did include exactly the same primer pairs as used in previous studies reporting positive results and optimized their working conditions for our laboratory. In addition, we used nested primer pairs, which amplified products in the same region of the viral genome and increased the sensitivity relative to the published primers by 10- to 100-fold. The nested PCR for RSV, in which the detection limit was established, proved to be extremely sensitive and could detect a fraction of a single infected cell. Similarly, it has previously been estimated that non-nested RT-PCR can detect 1000 copies of genomic RNA for MV, which is only a fraction of an average MV-infected cell, which has been shown to contain 30,000 copies of nucleocapsid mRNA and in addition genomic RNA.(58) With the nested PCR used here, the sensitivity could be increased 10-fold. It therefore seems unlikely that our technique would not have detected MV, even when considering that only few cells may be infected and that the level of viral transcription will be low in a persistent infection. The results from the ICC and ISH studies in particular, when seen together with the RT-PCR results, highlight some of the difficulties in interpreting data obtained with one technique only. Based on the ICC or ISH alone, we might have concluded that 1 patient showed presence of MV in bone marrow and 1 patient showed CDV. However, in both cases no confirmation was obtained on the same material using the second technique (ICC or ISH) and RT-PCR on part of the same biopsy specimen, tested exhaustively with all primer sets, was consistently negative also. The only tissue in which MV was found by both ICC and ISH (albeit weak staining) was in fact from a control patient with renal osteodystrophy. In this tissue also, no confirmation was obtained by RT-PCR. In 1 patient strong staining of the bone matrix was seen with the MV monoclonal antibody, suggesting that some cross-reactivity with an endogenous antigen may occur. Overall, we were unable to obtain consistent positive results with ISH and ICC (in the presence of appropriate positive and negative controls), which is in contrast with findings in diseases with a proven viral etiology. Together with the consistently negative data from the RT-PCR studies on bone biopsy specimens, bone marrow, and peripheral blood we conclude that the results from this large study do not support the belief that infection with a paramyxovirus is a necessary step in the pathogenesis of Paget's disease.


We are grateful for expert technical assistance in TEM from Mrs. L. Doverty and Mr. A. McKinnon (Aberdeen, U.K.). Mrs. P. Haddock and Dr. P. Duprex (Belfast, U.K.) assisted with ICC and ISH for MV. Dr. J. Compston (Cambridge, U.K.) and Dr. J. Kanis (Sheffield, U.K.) are thanked for help in obtaining bone biopsy specimens; Dr. J. Martin (Antwerp, Belgium) provided SSPE specimens for TEM, Dr. Musca Mircea (Cluj-Napoca, Romania) provided RSV-infected dog tissue, and Dr. B. Gimenez (Aberdeen, U.K.) provided RSV-infected cells. Mrs. G. Taylor and Mrs. I. Wilson (Aberdeen, U.K.) assisted with molecular techniques, and Dr. T. Pennington (Aberdeen, U.K.) made helpful suggestions during the course of these studies. This study was supported by grant C2074 from the Scottish Office Home and Health Department (M.H.H., R.P.H., and S.H.R).