REVIEWS Paraneoplastic syndromes in rheumatology Bernhard Manger and Georg Schett Abstract | For patients that present with musculoskeletal symptoms, diagnostic procedures carried out by physicians and rheumatologists are primarily aimed at confirming or excluding the occurrence of primary rheumatic diseases. Another important trigger for musculoskeletal disease, however, is the presence of a tumour. Careful clinical investigation and knowledge of the gestalt of musculoskeletal syndromes related to respective tumour entities is of utmost importance for the diagnosis of paraneoplastic rheumatic diseases such as hypertrophic osteoarthropathy, paraneoplastic polyarthritis, RS3PE syndrome, palmar fasciitis and polyarthritis, cancer-associated myositis and tumour-induced osteomalacia. This places great responsibility on rheumatologists in diagnosing malignancies and referring the patient for effective treatment. The selective influence of tumours on musculoskeletal tissue is surprising and indicates that tumours alter tissues such as the periosteum, synovial membrane, subcutaneous connective tissue, fascia, muscles and bones by specific molecular processes. Some of the underlying mechanisms have been unravelled, providing valuable information on the physiologic and pathophysiologic roles of mediators such as vascular endothelial growth factor and fibroblast growth factor 23. Manger, B. & Schett, G. Nat. Rev. Rheumatol. advance online publication 19 August 2014; doi:10.1038/nrrheum.2014.138

Introduction

Department of Internal Medicine 3, University of Erlangen-Nuremberg, Ulmenweg 18, D‑91054 Erlangen, Germany (B.M., G.S.). Correspondence to: B.M. bernhard.manger@ uk-erlangen.de

Paraneoplastic syndromes comprise diseases or symptoms that are not caused directly by the tumour or by its metastases, but are mediated by soluble factors, such as hormones and cytokines from a tumour, or are a consequence of humoral or cellular immune mechanisms directed against tumour cells and, therefore, often occur at regions distant from the underlying malignancy. In the case of paraneoplastic syndromes in rheumatology, which are rare, musculoskeletal symptoms arise in joints, fasciae, muscles, vessels or bones1 generally no longer than 2 years before the diagnosis of an associated neoplasm, and can therefore be of the utmost clinical importance for early detection and therapy. A variety of musculoskeletal symptoms can be associated with malignant diseases, but to consider a syndrome truly paraneoplastic, a causal relationship needs to be established. Almost 50 years ago, Sir Austin Bradford Hill proposed nine criteria to support an argument of causality over mere association (Box 1). Initially intended to judge the role of environmental factors in occupational medicine, these ‘Bradford Hill Criteria’ (also known as Hill’s criteria for causation) have since been used in many areas of medicine and can also be applied to para­neoplastic disorders. For this Review, we have selected those paraneoplastic syndromes that come closest to fulfiling these criteria of causality. In a paraneoplastic situation, the strongest argument is usually based on a close temporal relationship between the advent of musculo­skeletal symptoms and diagnosis of a tumour. The epidemio­logic evidence for representing a true paraneoplastic disorder Competing interests The authors declare no competing interests.

is strongest for HOA and CAM; by contrast, rare syndromes such as PFPAS or TIO exhibit the most specific clinical patterns. The ultimate proof of the para­neoplastic nature of a condition is provided by the fast regression of all symptoms following complete removal of tumour cells by surgery, chemotherapy or radiotherapy. Over the past two decades, some of the specific mechanisms that underlie paraneoplastic syndromes have been elucidated, which has added to our understanding of the pathogenesis of rheumatic as well as neoplastic diseases. This Review will focus on well-defined paraneoplastic musculoskeletal syndromes for which the evidence of a causal relationship to an underlying tumour is not only based on a temporal relationship but also on pathogenetic plausibility—hypertrophic osteoarthropathy, paraneoplastic polyarthritis, RS3PE syndrome, palmar fasciitis and polyarthritis, cancer-associated myositis and tumour-induced osteomalacia. It is also intended to enable rheumatologists to recognize the typical clinical patterns, to understand what is known about the underlying pathomechanisms and to draw the appropriate diagnostic and therapeutic conclusions.

Hypertrophic osteoarthropathy Clinical features The first description of periosteal thickening of long tubular bones in patients with benign chronic lung diseases dates back to the end of the nineteenth century, when the term ‘hypertrophic (pulmonary) osteo­arthro­ pathy’ (HOA) was created.2,3 It subsequently became clear that HOA can also present as a para­neoplastic dis­order, preceding other symptoms of a pulmonary malignancy.4,5 Tibial and femoral bone pain is the typical

NATURE REVIEWS | RHEUMATOLOGY

ADVANCE ONLINE PUBLICATION  |  1 © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS Key points ■■ Paraneoplasias are caused by soluble factors produced by tumour cells or are the consequence of an immune reaction against these cells ■■ An underlying malignancy should be sought if musculoskeletal symptoms do not show an adequate response to therapy ■■ Arthritis, myositis, periostitis, fasciitis and osteomalacia can all be musculoskeletal manifestations of a paraneoplastic disease ■■ Due to the properties of soluble factors produced by tumour cells, some paraneoplasias show a characteristic involvement of specific musculoskeletal tissues (e.g. VEGF inducing HOA or FGF23 inducing TIO) ■■ Despite their rarity, knowledge of distinct clinical patterns of paraneoplasias is essential for the rheumatologist, because their recognition allows for a timely diagnosis and potentially life-saving therapy ■■ The dissection of underlying molecular mechanisms can provide insights in the pathogenesis for rheumatic as well as neoplastic diseases

Box 1 | Bradford Hill criteria102 ■■ ■■ ■■ ■■ ■■ ■■ ■■ ■■ ■■

Strength of an association Consistency of findings by different researchers Specificity of a phenomenon Temporality or appropriate temporal sequence Biological gradient or dose-response effects Plausibility or pathophysiologic rationale Coherence with what is already known Experimental induction or abrogation of an effect Evidence from analogous conditions

musculoskeletal symptom associated with HOA, and arthralgia or synovitis of adjacent joints is common. Periosteal osseous proliferations can be detected by conventional radiography and the formation of new osseous tissue leads to increased tracer uptake in scintigraphic bone scans.6 The highly inflammatory character of perio­ stitis also generates a strong signal that can be detected by positron emission tomography.7 Another characteristic finding that, together with perio­stitis, bone and joint pain, constitutes HOA, is club­ bing of fingers and/or toes. This clinical sign, first des­ cribed by Hippocrates, is characterized by an increase in the angle of the hyponychium to over 180°, peri­ungual oedema and softening of the nail bed.8,9 In an analysis of over 1,200 bone scans from patients with lung cancer increased linear femoral or tibial tracer uptake was found in 4.5% of patients, but only 0.8% displayed the full clinical picture of HOA, with digital clubbing and arthralgias. 10 A very similar incidence rate of 0.72% from patients showing positive bone scintigram findings was confirmed in another study.11 A lesser-known clinical sign that can be found in association with HOA in patients with cancer is acanthosis palmaris (also known as tripe palms), a hyperkeratotic accentuation of dermato­glyphic lines, which gives the palmar skin a gyrated, velvety appearance (Figure 1).12,13

Pathogenesis HOA and clubbing are not only associated with pulmonary and other thoracic malignancies; that these traits are also associated with a variety of other hypoxaemic or inflammatory disorders has, for a long time, generated a number of pathogenic hypotheses,8,9,14 including the

suggestion in the initial paper on HOA that a soluble factor produced by the affected lung might influence peripheral tissues.3 One study found elevated serum concentrations of human growth hormone in patients with bronchial neoplasms and clubbing, 15 but this increase was not confirmed by others.16 Overproduction of a fibroblast growth factor was also proposed to be a common factor causing HOA and clubbing in many benign and malignant disorders.17 Platelet-derived growth factor In 1987, it was hypothesized that the release of plateletderived growth factor (PDGF) from small vessels of the fingertips in response to an accumulation of megakaryocyte fragments or platelet aggregates that have bypassed the lung capillary network in various cardiac and pulmonary diseases can increase vascularity, permeability and mesenchymal cell growth to promote new bone formation and clubbing.18 This hypothesis was supported by histopathologic findings of platelet microthrombi in clubbed fingers19 and high levels of von Willebrand factor antigen in the circulation of patients with HOA as a consequence of platelet or endothelial cell activation.20 Significantly increased concentrations of PDGF were reported in the sera of 21 patients with HOA in comparison to control patients with lung diseases but no HOA.21 Vascular endothelial growth factor PDGF is not the only growth factor involved in the pathogenesis of HOA, however. Immunohistochemical studies of clubbed fingertips have shown that vascular endothelial growth factor (VEGF) is upregulated in HOA to a greater extent than PDGF, and that both might synergize in its pathogenesis.22 Although hypoxaemia is a strong stimulus for the synthesis of VEGF, of all patients with secondary HOA, those with underlying malignancies have the highest plasma levels of VEGF.23,24 In one patient, surgical removal of a lung tumour resulted in the disappearance of all HOA-associated clinical symptoms coincident with normalization of VEGF concentrations; histochemical studies confirmed that the tumour tissue expressed abnormal levels of VEGF.25 Together with the known strong effect of VEGF on the osteoblastic differentiation of periosteal cells, these studies point to VEGF as a key cytokine in the pathogenesis of proliferative periostitis in HOA.26 Further insights into the pathogenesis of HOA come from investigation of the genetic background of pri­ mary HOA, a hereditary disease that closely mimics paraneoplastic HOA. Various mutations in HPGD, which encodes hydroxyprostaglandin dehydrogenase 15-(NAD), or SLCO2A1, which encodes solute carrier organic anion transporter family member 2A1, a prosta­ glandin transporter lead to a marked elevation in the levels of circulating prostaglandin E2 (PGE2).27,28 As it has long been known that PGE2 is a potent stimulus of VEGF-mediated osteoblast activation and bone formation, these findings add to our understanding of the pathogenesis of proliferative periostitis in HOA.29

2  |  ADVANCE ONLINE PUBLICATION

www.nature.com/nrrheum © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS Prognosis and therapy In the event of complete remission of the underlying malignancy achieved by resection, radio- or chemotherapy, clubbing and HOA can also completely diminish.30,31 Periostitis and bone pain usually respond well to prostaglandin inhibition by NSAIDs,6,32 which is consistent with the detection of mutations in enzymes of prostaglandin metabolism accompanied by high levels of circulating PGE2 in primary HOA, as outlined above. In cases when NSAIDs cannot sufficiently control the symptoms of HOA, zoledronic acid has been used successfully.33 In addition to directly suppressing bone turnover, zoledronic acid has also been shown to reduce VEGF levels in patients with metastatic tumours. 34 Another therapeutic alternative is the somatostatin analogue octreotide,35 which is presumably effective through its ability to inhibit VEGF production.36

Paraneoplastic polyarthritis Clinical features Cancer and polyarthritis are both common conditions in the general population, so their coexistence in one patient does not necessarily establish a causal relationship between the two conditions. Only when there is a close temporal relationship between the onset of poly­arthritis and the detection of a malignant process, or when successful tumour therapy leads to the remission of joint symptoms, does a paraneoplastic process seem likely.37 Within the past 30 years, eight case series analysing the clinical characteristics of a total of 121 patients in whom a malignancy was detected within less than 4 years after the onset of symptoms of arthritis have been published (Table 1).38–45 The median age of onset of arthritis symptoms was 54.2 years and the male:female ratio was 1.7:1, which clearly differs from the equivalent parameters for rheumatoid arthritis. About one third of all cases had a haematolymphatic malignancy, whereas the most frequently occurring solid tumours were adenocarcinomas of the lung and breast. The arthritis is usually of sudden onset, accompanied by high levels of inflammatory markers, such as C‑reactive protein and erythrocyte sedimentation rate. Of all patients with paraneoplastic arthritis, only 27.2% tested positive for rheumatoid factor and 19.0% for antinuclear antibodies.38–45 These data were confirmed by a study examining antibody profiles in patients with various paraneoplastic rheumatic syndromes.46 Evidence also exists that anti-citrullinated protein antibodies (ACPAs) can be present in patients with paraneoplastic arthritis,47,48 with one case series reporting ACPA positivity in 7 of 65 patients (10.7%).45 Pathogenesis No convincing pathogenetic hypothesis for paraneoplastic arthritis has so far been proposed. Earlier claims that circulating immune complexes were involved could not be substantiated.49 In one patient with renalcell carcinoma and oligoarthritis, a clone with identical T‑cell-receptor γ gene rearrangement was isolated from synovial tissue as well as from tumour-infiltrating cells, suggesting that lymphocytes directed against the

Figure 1 | Acanthosis palmaris or tripe palms in a patient with lung cancer who had also developed marked clubbing of the terminal phalanges of all fingers. Permission obtained from C. Amberger, Private practice, Bad Neuenahr, Germany.

tumour can crossreact with synovial antigens to trigger p­araneoplastic synovitis.50

Prognosis and therapy A characteristic feature of paraneoplastic arthritis is a rather poor response to NSAIDs and steroid therapy in comparison with other forms of early arthritis.42,45 Remis­ sion is usually not achieved by immunosuppression, but successful surgical removal or chemotherapy of the underlying malignancy frequently leads to the complete resolution of all rheumatic symptoms.42–44 In most cases, however, a tumour relapse is not accompanied by the recurrence of arthritis.42,44

RS3PE Clinical features Remitting seronegative symmetrical synovitis with pit­ ting oedema (RS3PE), a form of polyarthritis that occurs in elderly people, was first described by McCarty et al. in 1985 (Figure 2).51 It is characterized by the symmetrical involvement of small joints and marked pitting oedema on the dorsum of the hands and feet, a sudden inflammatory onset, rheumatoid factor negativity and an overall excellent prognosis. 52 Soon after the initial description of this form of polyarthritis, increasing numbers of cases were reported, with a neoplastic process being diagnosed immediately after the first symptoms of RS3PE or during follow-up.52,53 In five small case series of a total of 89 patients with RS3PE, a malignancy was reported in 22 patients (24.7%), in five cases of haemato­ poietic origin.54–58 No significant demographic or clinical differences were observed between idiopathic and paraneoplast­ic cases of RS3PE. Pathogenesis VEGF has been proposed to be important in the pathogenesis of RS3PE in both idiopathic and paraneoplastic cases, with elevated serum levels in comparison with

NATURE REVIEWS | RHEUMATOLOGY

ADVANCE ONLINE PUBLICATION  |  3 © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS Table 1 | Demographic and clinical characteristics of patients with paraneoplastic arthritis* Study

Number of patients

Males:females

Mean age (years)

Tumour type (haematologic: solid)

Arthritis type (polyarthritis: oligoarthritis)

RFpositivity (%)

Time between onset of arthritis and tumour diagnosis (months)

Pines et al.38

3

1:2

63.3

0:3

2:1

67

3.0

Alvarez Lario et al.39

5

2:3

65.4

0:5

4:5

40

4.2

Pfitzenmeyer et al.40

12

7:5

61.2

0:12

12:0

42

3.3

Stummvoll et al.41

2

2:0

59.5

0:2

2:0

0

8.0

Morel et al.42

26

16:10

57.5

6:20

22:4

31

4.4

Hakkou et al.43

3

2:1

34.3

3:0

3:0

0

4.3

Yamashita et al.44

5

3:2

65.8

5:0

4:1

20

19.2

Kisacik et al.45

65

43:22

50.2

26:39

22:31 (12 with monoarthritis)

23

5.1

*Published in case series since 1984. Abbreviation: RF, rheumatoid factor.

carcinomas might also respond to high doses of glucocorticoids, it is recommended that RS3PE is treated using only low doses of prednisone (≤10 mg/d) to more effectively differentiate from an underlying malignancy.54 Similar to the other forms of arthritis discussed earlier, only the prompt and sustained remis­sion of all rheumatic symptoms after the successful removal of a malignant tumour can confirm the paraneoplastic nature of this condition.66,67

Palmar fasciitis and polyarthritis Figure 2 | Remitting seronegative symmetrical synovitis with pitting oedema in a patient with non-Hodgkin lymphoma.

other autoimmune rheumatic diseases reported for idio­ pathic RS3PE as well as for cases associated with sarcoidosis or angioimmunoblastic T‑cell lymphoma.59–61 Since VEGF is a potent angiogenic and vasoactive molecule, it could facilitate both synovial hypervascularity (synovitis) and increased vascular permeability (subcutaneous oedema) in RS3PE.59 So far, the only marker specific for paraneoplastic RS3PE is matrix metallo­ proteinase (MMP)‑3, the levels of which are significantly elevated in the serum.58 The pathogenetic relevance of this finding remains unclear, but MMPs have been shown to be involved in the invasion or progression of solid tumours as well as in the pathologic destruction of joint tissues in arthritides.62,63

Prognosis and therapy Idiopathic RS3PE shows an excellent and rapid response to glucocorticoid therapy. This is also true for many para­ neoplastic RS3PE cases,52,57,58 although sometimes the steroid response appears to be diminished or delayed.54,64,65 Because haematopoietic neoplasms and some metastatic

Clinical features Palmar fasciitis and polyarthritis syndrome (PFPAS) is a rare paraneoplastic disorder first described as a separate entity in 1982 by Medsger et al.,68 who reported the sudden onset of stiffness and diffuse painful swelling of both hands together with polyarthritis in six postmenopausal women with malignant ovarian tumours. Previously, similar symptoms occurring in patients with tumours had often been classified as para­neoplastic shoulder hand syndrome or reflex sympathetic dys­ trophy.69 A comprehensive review published in 2014 has analysed the clinical and laboratory data of 100 patients with this syndrome described in the literature.70 The hallmark of PFAPS is inflammation of the palmar fascia, which leads to flexion contractures with nodular thickening similar to, but more severe than, that occurring in Dupuytren’s contracture (Figure 3). The palpatory sensation of the marked induration of subcutaneous tissues has led to the descriptive term ‘woody hands’.71,72 The poly­ arthritis aspect involves meto­carpophalangeal and proximal interphalangeal joints and wrists, while arthritides of other joints are frequent but usually milder. Carpal tunnel syndrome can be present. The most frequent tumour type in patients with paraneoplastic PFPAS is ovarian adenocarcinoma (present in 37% of cases), which, together with breast cancer

4  |  ADVANCE ONLINE PUBLICATION

www.nature.com/nrrheum © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS and other malignancies involving female reproductive organs, is responsible for more than half of all published cases of paraneoplastic PFPAS.70 For this reason, PFPAS is over four times more prevalent in females than in males. In patients with PFPAS, inflammation markers show a great range of variability, rheumatoid factor is negative or only weakly positive, and ACPAs have not been detected. Most patients test positive for at least one tumour marker, most frequently CA125 or CA19‑9.70

Pathogenesis The pathogenesis of this paraneoplastic syndrome is still unknown. The predominance of females with PFPAS has often been used as an argument for a patho­physiologic role for female sex hormones, but evidence to support this theory is lacking. Fibroblast proliferation and increased production of extracellular matrix components are key histologic features of PFPAS, suggesting a potential role for tumour-derived soluble stimulators of fibroblast activity; significantly increased levels of connective tissue growth factor have been found in the serum of one patient with PFPAS, but this observation remains to be confirmed in others.73 One patient with gastro-oesophageal carcinoma who received an experimental treatment with a broad-spectrum MMP inhibitor for adjuvant chemo­ therapy developed PFPAS symptoms during treatment; these symptoms disappeared when the therapy was interrupted but returned following its reintro­duction, suggesting that a reduced activity of certain MMPs may be crucial for the pathogenesis of PFPAS.74 Lymphocyte infiltration is a regular histologic feature of the affected tissues, and the detection of IgG deposits in some cases suggests a role for immune mechanisms in this syndrome.75,76 Prognosis and therapy The treatment of PFPAS with NSAIDs is usually not effective, and the effect of glucocorticoids—even at high doses—is disappointing. No immunosuppressive agent has yet shown any therapeutic benefit.70 The prognosis of patients with PFPAS is dismal because the malignancy has usually already advanced to a late metastatic stage by the time the symptoms of PFPAS manifest. In rare cases, however, when complete surgical resection of the tumour is possible, all musculoskeletal symptoms can undergo complete remission.77–78

Cancer-associated myositis Clinical features The first description of cancer-associated myositis (CAM) was confirmed from a biopsy sample taken from a patient with gastric adenocarcinoma almost 100 years ago.79 Numerous epidemiologic studies have since studied the relationship between various types of inflammatory muscle disease and malignancies. A meta-analysis of six population-based cohort studies revealed the highest association for a close temporal relationship between tumours and dermatomyositis, with standard incidence ratios (SIRs) ranging from 3.8 to 7.7.80 The association is much weaker for polymyositis (the SIR in five studies ranged from 1.7 to 2.15) or inclusion-body myositis,

Figure 3 | Nodular palmar fasciitis and flexion contractures in a patient with metastatic ovarian carcinoma. The skin changes at the tip of the index finger date back to an earlier injury and are not related to the paraneoplastic disease.

and some studies have failed to detect any relationship at all for these types.80–82 Malignancies have also been described in association with amyopathic forms of dermato­myositis.83 The most frequently occurring can­ cers in patients with dermatomyositis are ovarian, pancreatic, stomach and colorectal, whereas patients with poly­myositis often have lung and bladder cancers; nonHodgkin lymphoma shows an association with both diseases.84 In Asian populations, the most frequent tumour type is nasopharyngeal carcinoma.85 Patients with CAM tend to be older and show a more severe muscular and cutaneous involvement (with skin ulcerations), but a lower incidence of interstitial lung dis­ ease and anti-Jo‑1 autoantibody positivity, than patients with other forms of myositis.86,87 Indeed, the complete absence of reactivity shown in a routinely assessed myositis-associated autoantibody panel (against Jo‑1, PM‑Scl, U1-RNP, U3-RNP and Ku) has a positive predictive value for the presence of an underlying malignancy in myositis.87

Pathogenesis Somewhat at odds with this positive predictive value in the absence of reactivity, pathophysiologic studies by Casciola-Rosen et al.88 demonstrated high expression levels of the myositis-specific autoantigens Mi‑2, Jo‑1 and Ku in inflamed muscle tissue, regenerating muscle cells, cultured myoblasts, and various tumour cell types. The authors propose that in CAM, the immune response against cancer cells could crossreact with regenerating muscle tissue in myositis to perpetuate inflammation.88 This hypothesis is supported by a report providing histochemical evidence for subclinical myopathy in clinically healthy muscle tissue of patients with colorectal cancer.89 In 2006, an autoantibody directed against a 155 kDa intracellular protein (anti‑p155) was found in the sera

NATURE REVIEWS | RHEUMATOLOGY

ADVANCE ONLINE PUBLICATION  |  5 © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS Hypertrophic osteoarthropathy 6

1 Periosteal proliferation

Osteomalacia

Polyarthritis

VEGF, PDGF

Phosphaturia

Synovitis Bronchial

FGF23

2

?

Mesenchymal

Various*

Various*

Various*

5

TIF1

Ovarian

MMP-3 Oedema

Myositis CTGF?

CAM

R3SPE

Fasciitis 4

3 PFPAS

Figure 4 | Pathophysiology and clinical signs of paraneoplastic rheumatic syndromes. The circles emanating from the centre indicate the predominant type of tumour and the molecular mechanism responsible for the musculoskeletal disease as well as the pathologic process of the respective paraneoplastic syndrome. The cartoons indicate the various processes involved: (1) proliferation (red) of periosteal bone lining cells (purple) in hypertrophic osteoarthropathy; (2) inflammatory arthritis with synovitis in paraneoplastic polyarthritis; (3) diffuse hand swelling and pitting oedema (arrow) in remitting seronegative symmetrical synovitis with pitting oedema; (4) cell infiltration and inflammation of the facia (orange) between the muscles (brown) in palmar fasciitis and polyarthritis; (5) cellular infiltration of the muscle in paraneoplastic myositis and (6) bone loss (bone: white, bone marrow: grey) in parneoplastic osteomalacia. *Haematolymphatic and solid tumours such as breast, prostate, lung and gastrointestinal cancers. Abbreviations: CAM, cancer-associated myositis; CTGF, connective tissue growth factor; FGF23, fibroblast growth factor 23; MMP3, matrix metalloproteinase 3; PDGF, platelet-derived growth factor; PFPAS, palmar fasciitis and polyarthritis; R3SPE, remitting seronegative symmetrical synovitis with pitting oedema; TIF1, transcription intermediary factor 1; VEGF, vascular endothelial growth factor.

of patients with juvenile and adult dermatomyositis but not in those with polymyositis.90 It subsequently became clear that many of the adult patients showing p155 positivity had CAM; some also expressed antibodies against an associated 140 kDa protein (anti‑p155/140). Of 445 adult patients with dermatomyositis, 74 expressed anti‑p155 or anti‑p155/140 antibodies, and in 64.9% a malignancy was detected within 3 years of diagnosis].91 A meta-analysis of six other cohorts of patients with myo­sitis found a sensitivity of 78% and specificity of 89% for anti‑p155 antibodies in the diagnosis of CAM.92 Further characterization of anti‑p155 and anti‑p140 revealed their targets to be transcription intermediary factor (TIF)‑1γ and TIF‑1α, respectively,91 which belong to a protein family with intriguing roles in carcino­ genesis. For example, TIF‑1α ubiquitinates the tumour suppressor protein p53 to negatively regulate its levels, so

TIF‑1α depletion causes the p53-dependent, spontaneous apoptosis of human breast cancer cells.93 Thus, an anti-tumour immune response could induce anti-TIF auto­antibodies that contribute to the patho­genesis of CAM, by crossreactivity of the immune response against tumour cells with regenerating muscle tissue. 91 The detec­tion of TIF‑1γ autoantibodies can now reliably be per­formed by immunoblot or ELISA assays in a standard laboratory setting.94 Another autoantibody, designated anti-MJ, which binds the nuclear matrix protein NXP2, might also have the potential to distinguish patients with CAM, but preliminary data suggesting an association of this autoantibody with CAM still need to be confirmed in larger cohorts.95

Prognosis and therapy CAM usually responds well to glucocorticoid therapy, although skin manifestations can be recalcitrant. In a retro­spective study, the 5‑year survival rate of patients with CAM was considerably worse (56%) than that of patients with idiopathic forms of myositis (92%). Of those patients with CAM, 53% went into remission after surgical, radiotherapeutic or chemotherapeutic remo­val of the malignancy. In some cases, however, the symp­toms of myositis recurred even without a relapse of the can­cer, supporting the hypothesis that an initially tumour-triggered, but later self-perpetuating, immune response occurs against muscle antigens.86

Tumour-induced osteomalacia Clinical features In 1947, McCance described the case of a 15‑year-old girl with bone pain, gait disturbances and low levels of serum phosphorus that did not respond to high doses of vita­ min D; only after the removal of a bone tumour from her femur did her symptoms completely resolve.96 More than 300 similar cases have subsequently been reported in the literature, and the syndrome has been named tumourinduced osteomalacia (TIO) or oncogenic osteo­malacia. It is characterized by pathologic frac­tures, muscle weak­ness, height loss, hypo­phos­phataemia, hyperphosphaturia, and normal or low levels of 1,25‑­dihydroxyvitamin D.97 Other causes of genetic and acquired hypophosphataemia must be excluded in the diagnostic work-up and, in most cases, it often takes a long time to identify and locate the un­derlying tumour. Pathogenesis The idea that a circulating factor can cause phosphate wasting is not novel,98 but the breakthrough in unravelling the pathogenesis of TIO came with the discovery that tumours secrete the endocrine fibroblast growth factor 23 (FGF23; also known as phosphatonin), a member of the fibroblast growth factor superfamily, which binds to proximal tubule cells of the kidney and thereby induces a marked increase in phosphate excretion.97,99 The typical type of neoplasm associated with TIO and the production of FGF23 is a phosphaturic mesenchymal tumour (mixed connective tissue variant), but other histologic diagnoses, such as haemangiopericytoma, osteosarcoma,

6  |  ADVANCE ONLINE PUBLICATION

www.nature.com/nrrheum © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS giant cell tumour and other mesenchymal tumours, have also been described.100 In a cohort of 39 new patients and a review of 269 published cases of TIO, neoplasms—only 8% of which were malignant—were found to originate predominantly in the bone in 40% of cases and in soft tissue in 55% of cases; the lower extremities were the most frequent location.101 As the tumours are often small, their detection can be challenging, but octreotide scinti­ graphy or positron emission tomography have often been successfully used.97,101 A group of hereditary diseases is also associated with a markedly increased production of FGF23 and produces very similar effects on target organs, but these symptoms usually manifest during childhood, leading to the clinical picture of rickets (these diseases were formerly known as vitamin D‑resistant rickets); by contrast, TIO is uncommon before the age of 40 and resembles adult osteomalacia of other causes.101

Prognosis and therapy The prognosis for TIO is excellent following complete resection of the occult neoplasm, which leads to the rapid and complete reversal of all symptoms. A return of serum FGF23 levels to normal indicates the successful removal of the entire tumour. If the tumour cannot be found or its removal is not possible, medical therapy using phosphate supplementation and active vitamin D compounds is necessary.97

Review criteria

Conclusions

In this Review, we have tried to summarize the plausibility and coherence of pathophysiologic mechanisms des­cribed for a number of paraneoplastic disorders in rheumatology (Figure 4). Some other rheumatic disorders that have also occasionally been described in association with malignancies, such as adult-onset Still’s disease or polymyalgia rheumatica, have not been included on the basis of them not achieving the Bradford-Hill criteria. Despite most paraneoplastic syndromes in rheumatology being rare, they are of major clinical importance because they often precede other clinical manifestations 1.

2.

3.

4.

5.

6.

7.

Azar, L. & Khasnis, A. Paraneoplastic rheumatologic syndromes. Curr. Opin. Rheumatol. 25, 44–49 (2013). Von Bamberger, E. Veränderungen der Röhrenknochen bei Bronchiektasie [German]. Wien. Klin. Wochenschr. 2, 226–240 (1889). Marie, P. De l´ostéo-arthropathie hypertrophiante pneumonique [French]. Rev. Med. Paris 10, 1–36 (1890). Craig, J. W. Hypertrophic pulmonary osteoarthropathy as the first symptom of pulmonary neoplasm. Brit. Med. J. 1, 750–752 (1937). Vogl, A., Blumenfeld, S. & Gutner, L. B. Diagnostic significance of pulmonary hypertrophic osteoarthropathy. Am. J. Med. 18, 51–65 (1955). Pineda, C. & Martínez-Lavín, M. Hypertrophic osteoarthropathy: what a rheumatologist should know about this uncommon condition. Rheum. Dis. Clin. North Am. 39, 383–400 (2013). Manger, B. et al. Clinical Images: Hippokrates confirmed by positron emission tomography. Arthritis Rheum. 63, 1150 (2011).

of neoplasms and can facilitate the timely diagnosis and potential cure of a malignant disease. If the malignant cells can be successfully eliminated, the paraneo­ plastic symptoms usually subside, but the reappearance of musculo­skeletal symptoms can indicate a relapse or metastatic spreading, although clinically this is not a reliable phenomenon. Paraneoplastic symptoms can also have a significant impact on the quality of life, morbidity and mortality of patients with tumours. Scientifically, paraneoplasias provide an interesting link between the pathomechanisms of neoplastic and rheumatic disorders, which still hold a variety of unresolved questions. One of the greatest puzzles in the field of para­neo­ plasias is the ‘homing’ of certain disease manifestations to specific tissues, organs or locations. When soluble mediators produced by tumour cells are responsible for inducing a defined pathology such as periostitis or fasciitis, no satisfactory explanation currently exists for the preferential location at the palmar fascia in PFPAS or the tibial shaft in HOA, for example. Further research to address these questions is required, although it is hampered by the rarity of these diseases. International registries for paraneoplasias in rheumatology with the collection of biologic materials could help to overcome those difficulties and increase our knowledge about this group of rare but important diseases.

8.

9. 10.

11.

12.

13.

14.

We searched for articles focusing on paraneoplastic syndromes in rheumatology in PubMed published from January 1950 to April 2014 using the search terms “hypertrophic osteoarthropathy and cancer”, “palmar fasciitis”, “RS3PE”, “tumour-induced osteomalacia”, or “oncogenic osteomalacia” or “paraneopl* and arthritis” or “paraneopl* and myositis”. All articles identified as relevant for this Review because they discussed pathogenetic aspects of these diseases were reviewed as full-length papers in English, French, German or Spanish language. In addition, we searched the reference lists of selected articles for further relevant publications.

Martínez-Lavín, M. Exploring the cause of the most ancient clinical sign in medicine: finger clubbing. Semin. Arthritis Rheum. 36, 380–385 (2007). Rutherford, J. D. Digital clubbing. Circulation 127, 1997–1999 (2013). Izumi, M., Takayama, K., Yabuuchi, H., Abe, K. & Nakanishi, Y. Incidence of hypertrophic pulmonary osteoarthropathy associated with primary lung cancer. Respirology 15, 809–812 (2010). Ito, T. et al. Hypertrophic pulmonary osteoarthropathy as a paraneoplastic manifestation of lung cancer. J. Thorac. Oncol. 5, 976–980 (2010). Cohen, P. R. Hypertrophic pulmonary osteoarthropathy and tripe palms in a man with squamous cell carcinoma of the larynx and lung. Am. J. Clin. Oncol. 16, 268–276 (1993). Saeed, H. & Massarweh, S. Images in clinical medicine. Hypertrophic pulmonary osteoarthropathy and tripe palms. N. Engl. J. Med. 366, 360 (2012). Spicknall, K. E., Zirwas, M. J. & English III, J. C. Clubbing: An update on

NATURE REVIEWS | RHEUMATOLOGY

diagnosis, pathophysiology, and clinical relevance. J. Am. Acad. Dermatol. 52, 1020–1028 (2005). 15. Gosney, M. A., Gosney, J. R. & Lye, M. Plasma growth hormone and digital clubbing in carcinoma of the bronchus. Thorax 45, 545–547 (1990). 16. Yorgancioğlu, A., Akin, M., Demtray, M. & Derelt, S. The relationship between digital clubbing and serum growth hormone level in patients with lung cancer. Monaldi Arch. Chest Dis. 51, 185–187 (1996). 17. Martínez-Lavín, M. Digital clubbing and hypertrophic osteoarthropathy: a unifying hypothesis. J. Rheumatol. 14, 6–8 (1987). 18. Dickinson, C. J. & Martin, J. F. Megakaryocytes and platelet clumps as the cause of finger clubbing. Lancet 2, 1434–1435 (1987). 19. Fox, S. B., Day, C. A. & Gatter, K. C. Association between platelet microthrombi and finger clubbing. Lancet 338, 313–314 (1991). 20. Matucci-Cerinic, M., Martinez-Lavin, M., Rojo, F., Fonseca, C. & Kahaleh, B. M. Von Willebrand factor antigen in hypertrophic osteoarthropathy. J. Rheumatol. 19, 765–767 (1992).

ADVANCE ONLINE PUBLICATION  |  7 © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS 21. Silveri, E. et al. Hypertrophic osteoarthropathy: endothelium and platelet function. Clin. Rheumatol. 15, 435–439 (1996). 22. Atkinson, S. & Fox, S. B. Vascular endothelial growth factor (VEGF)‑A and platelet-derived growth factor (PDGF) play a central role in the pathogenesis of digital clubbing. J. Pathol. 203, 721–728 (2004). 23. Silveira, L. H. et al. Vascular endothelial growth factor and hypertrophic osteoarthropathy. Clin. Exp. Rheumatol. 18, 57–62 (2000). 24. Abe, Y. et al. A case of pulmonary adenocarcinoma associated with hypertrophic osteoarthropathy due to vascular endothelial growth factor. Anticancer Res. 22, 3485–3488 (2002). 25. Olán, F. et al. Circulating vascular endothelial growth factor concentrations in a case of pulmonary hypertrophic osteoarthropathy. J. Rheumatol. 31, 614–616 (2004). 26. Hah, Y. S. et al. Vascular endothelial growth factor stimulates osteoblastic differentiation of cultured human periosteal-derived cells expressing vascular endothelial growth factor receptors. Mol. Biol. Rep. 38, 1443–1450 (2011). 27. Uppal, S. et al. Mutations in 15hydroxyprostaglandin dehydrogenase cause primary hypertrophic osteoarthropathy. Nat. Genet. 40, 789–793 (2008). 28. Zhang, Z., He, J. W., Fu, W. Z., Zhang, C. Q. & Zhang, Z. L. A novel mutation in the SLCO2A1 gene in a Chinese family with primary hypertrophic osteoarthropathy. Gene 521, 191–194 (2013). 29. Harada, S. et al. Induction of vascular endothelial growth factor expression by prostaglandin E2 and E1 in osteoblasts. J. Clin. Invest. 93, 2490–2496 (1994). 30. Shih, W. J. Pulmonary hypertrophic osteoarthropathy and its resolution. Semin. Nucl. Med. 34, 159–163 (2004). 31. Ulusakarya, A. et al. Symptoms in cancer patients and an unusual tumor: Case 1. Regression of hypertrophic pulmonary osteoarthropathy following chemotherapy for lung metastases of a nasopharyngeal carcinoma. J. Clin. Oncol. 23, 9422–9423 (2005). 32. Kozak, K. R., Milne, G. L., Morrow, J. D. & Cuiffo, B. P. Hypertrophic osteoarthropathy pathogenesis: a case highlighting the potential role for cyclo‑oxygenase‑2-derived prostaglandin E2. Nat. Clin. Pract. Rheumatol. 2, 452–456 (2006). 33. Johnson, S. A., Spiller, P. A. & Faull, C. M. Treatment of resistant pain in hypertrophic pulmonary osteoarthropathy with subcutaneous octreotide. Thorax 52, 298–299 (1997). 34. Angel-Moreno Maroto, A., Martínez-Quintana, E., Suárez-Castellano, L. & Pérez-Arellano, J. L. Painful hypertrophic osteoarthropathy successfully treated with octreotide. The pathogenetic role of vascular endothelial growth factor (VEGF). Rheumatology 44, 1326–1327 (2005). 35. Jayakar, B. A., Abelson, A. G. & Yao, Q. Treatment of hypertrophic osteoarthropathy with zoledronic acid: case report and review of the literature. Semin. Arthritis Rheum. 41, 291–296 (2011). 36. Santini, D. et al. Changes in bone resorption and vascular endothelial growth factor after a single zoledronic acid infusion in cancer patients with bone metastases from solid tumours. Oncol. Rep. 15, 1351–1357 (2006). 37. Naschitz, J. E. & Rosner, I. Musculoskeletal syndromes associated with malignancy

(excluding hypertrophic osteoarthropathy). Curr. Opin. Rheumatol. 20, 100–105 (2008). 38. Pines, A., Kaplinsky, N., Olchovsky, D. & Frankl, O. Rheumatoid arthritis-like syndrome: A presenting symptom of malignancy. Report of 3 cases and review of the literature. Eur. J. Rheum. Inflam. 7, 51–55 (1984). 39. Alvarez Lario, B. et al. Poliartritis paraneoplásica. Descripción de cinco casos [Spanish]. Med. Clin. (Barc.) 88, 55–58 (1987). 40. Pfitzenmeyer, P., Bielefeld, P., Tavernier, C., Besancenot, J. F. & Gaudet, M. Aspects actuels de la polyarthrite aiguë paranéoplasique [French]. Rev. Med. Interne 13, 195–199 (1992). 41. Stummvoll, G. H., Aringer, M. Machold, K. P., Smolen, J. S. & Raderer, M. Cancer polyarthritis resembling rheumatoid arthritis as a first sign of hidden neoplasms. Scand. J. Rheumatol. 30, 40–44 (2001). 42. Morel, J. et al. Characteristics and survival of 26 patients with paraneoplastic arthritis. Ann. Rheum. Dis. 67, 244–247 (2008). 43. Hakkou, J., Rostom, S., Bahiri, R. & HajjajHassouni, N. Paraneoplastic syndromes: report of eight cases and review of literature. Rheumatol. Int. 32, 1485–1489 (2012). 44. Yamashita, H. et al. Characteristics of 10 patients with paraneoplastic rheumatologic musculoskeletal manifestations. Mod. Rheumatol. 24, 492‑ 498 (2014). 45. Kisacik, B. et al. Diagnostic dilemma of paraneoplastic arthritis: case series. Int. J. Rheum. Dis. http://dx.doi.org/10.1111/ 1756-185X.12277. 46. Rugienė, R. et al. Prevalence of paraneoplastic rheumatic syndromes and their antibody profile among patients with solid tumours. Clin. Rheumatol. 30, 373–380 (2011). 47. Kumar, S., Sethi, S., Irani, F. & Bode, B. Y. Anticyclic citrullinated peptide antibody-positive paraneoplastic polyarthritis in a patient with metastatic pancreatic cancer. Am. J. Med. Sci. 338, 511–512 (2009). 48. Larson, E., Etwaru, D., Siva, C. & Lawlor, K. Report of anti-CCP antibody positive paraneoplastic polyarthritis and review of the literature. Rheumatol. Int. 31, 1635–1638 (2011). 49. Bradley, J. D. & Pinals, R. S. Carcinoma polyarthritis: role of immune complexes in pathogenesis. J. Rheumatol. 10, 826–828 (1983). 50. Schultz, H., Krenn, V. & Tony, H. P. Oligoarthritis mediated by tumor-specific T lymphocytes in renal-cell carcinoma. N. Engl. J. Med. 341, 290–291 (1999). 51. McCarty, D. J., O’Duffy, J. D., Pearson, L. & Hunter, J. B. Remitting seronegative symmetrical synovitis with pitting edema. RS3PE syndrome. JAMA 254, 2763–2767 (1985). 52. Yao, Q., Su, X. & Altman, R. D. Is remitting seronegative symmetrical synovitis with pitting edema (RS3PE) a subset of rheumatoid arthritis? Semin. Arthritis Rheum. 40, 89–94 (2010). 53. Sibilia, J. et al. Remitting seronegative symmetrical synovitis with pitting edema (RS3PE): a form of paraneoplastic polyarthritis? J. Rheumatol. 26, 115–120 (1999). 54. Paira, S., Graf, C., Roverano, S. & Rossini, J. Remitting seronegative symmetrical synovitis with pitting oedema: a study of 12 cases. Clin. Rheumatol. 21, 146–149 (2002). 55. Russell, E. B. Remitting seronegative symmetrical synovitis with pitting edema syndrome: followup for neoplasia. J. Rheumatol. 32, 1760–1761 (2005).

8  |  ADVANCE ONLINE PUBLICATION

56. Fietta, P. & Manganelli, P. Remitting seronegative symmetrical synovitis with pitting edema syndrome: followup for neoplasia. J. Rheumatol. 33, 2365–2366 (2006). 57. Bucaloiu, I. D., Olenginski, T. P. & Harrington, T. M. Remitting seronegative symmetrical synovitis with pitting edema syndrome in a rural tertiary care practice: a retrospective analysis. Mayo Clin. Proc. 82, 1510–1515 (2007). 58. Origuchi, T. et al. High serum matrix metalloproteinase 3 is characteristic of patients with paraneoplastic remitting seronegative symmetrical synovitis with pitting edema syndrome. Mod. Rheumatol. 22, 584–588 (2012). 59. Arima, K. et al. RS3PE syndrome presenting as vascular endothelial growth factor associated disorder. Ann. Rheum. Dis. 64, 1653–1655 (2005). 60. Matsuda, M. et al. Sarcoidosis with high serum levels of vascular endothelial growth factor (VEGF), showing RS3PE-like symptoms in extremities. Clin. Rheumatol. 23, 246–248 (2004). 61. Tabeya, T. et al. A case of angioimmunoblastic T‑cell lymphoma with high serum VEGF preceded by RS3PE syndrome. Mod. Rheumatol. http:// dx.doi.org/10.3109/14397595.2013.857836. 62. Zucker, S, Vacirca, J. Role of matrix metalloproteinases (MMPs) in colorectal cancer. Cancer Metastasis Rev. 23, 101–117 (2004). 63. Murphy, G. & Nagase, H. Reappraising metalloproteinases in rheumatoid arthritis and osteoarthritis: destruction or repair? Nat. Clin. Pract. Rheumatol. 4, 128–135 (2007). 64. Roldan, M. R., Martinez, F., Roman, J. & Torres, A. Non-Hodgkin’s lymphoma: initial manifestation. Ann. Rheum. Dis. 52, 85–86 (1993). 65. Chiappetta, N. & Gruber, B. Remitting seronegative symmetrical synovitis with pitting edema associated with acute myeloid leukemia. J. Rheumatol. 32, 1613–1614 (2004). 66. Tada, Y. et al. Remitting seronegative symmetrical synovitis with pitting edema associated with gastric carcinoma. J. Rheumatol. 24, 974–975 (1997). 67. Olivo, D. & Mattace, R. Concurrence of benign edematous polysynovitis in the elderly (RS3PE syndrome) and endometrial adenocarcinoma. Scand. J. Rheumatol. 26, 67–68 (1997). 68. Medsger, T. A., Dixon, J. A. & Garwood, V. F. Palmar fasciitis and polyarthritis associated with ovarian carcinoma. Ann. Intern. Med. 96, 424–431 (1982). 69. Bremer, C. Shoulder-hand syndrome. A case of unusual aetiology. Ann. Phys. Med. 9, 168–171 (1967). 70. Manger, B. & Schett, G. Palmar fasciitis and polyarthritis syndrome—systematic literature review of 100 cases. Semin. Arthritis Rheum. http://dx.doi.org/10.1016/j.semarthrit.2014. 03.005. 71. Cox, N. H., Ramsay, B., Dobson, C. & Comaish, J. S. Woody hands in a patient with pancreatic carcinoma: a variant of cancerassociated fasciitis-panniculitis syndrome. Br. J. Dermatol. 135, 995–998 (1996). 72. Alexandroff, A. B. et al. Woody hands. Lancet 361, 1344 (2003). 73. Yogarajah, M., Soh, J., Lord, B., Goddard, N. & Stratton, R. Palmar fasciitis and polyarthritis syndrome: a sign of ovarian malignancy. J. R. Soc. Med. 101, 473–475 (2008). 74. Virik, K., Lynch, K. P. & Harper, P. Gastroesophageal cancer, palmar fasciitis and a matrix metalloproteinase inhibitor. Intern. Med. J. 32, 50–51 (2002).

www.nature.com/nrrheum © 2014 Macmillan Publishers Limited. All rights reserved

REVIEWS 75. Shiel, W. C. Jr., Prete, P. E., Jason, M. & Andrews, B. S. Palmar fasciitis and arthritis with ovarian and non-ovarian carcinomas. New syndrome. Am. J. Med. 79, 640–644 (1985). 76. Valverde-Garcia, J. et al. Paraneoplastic palmar fasciitis-polyarthritis syndrome associated with breast cancer. J. Rheumatol. 14, 1207–1209 (1987). 77. Goldberg, E., Dobransky, R. & Gill, R. Reflex sympathetic dystrophy associated with malignancy. Arthritis Rheum. 28, 1079–1080 (1985). 78. Enomoto, M. et al. Palmar fasciitis and polyarthritis associated with gastric carcinoma: complete resolution after total gastrectomy. Intern. Med. 39, 754–757 (2000). 79. Stertz, G. Polymyositis [German]. Berl. Klin. Wochenschr. 53, 489 (1916). 80. Zahr, Z. A. & Baer, A. N. Malignancy in myositis. Curr. Rheumatol. Rep. 13, 208–215 (2011). 81. Airio, A., Pukkala, E. & Isomäki, H. Elevated cancer incidence in patients with dermatomyositis: a population based study. J. Rheumatol. 22, 1300–1303 (1995). 82. Limaye, V. et al. The incidence and associations of malignancy in a large cohort of patients with biopsy-determined idiopathic inflammatory myositis. Rheumatol. Int. 33, 965–971 (2013). 83. Whitmore, S. E., Watson, R., Rosenshein, N. B. & Provost, T. T. Dermatomyositis sine myositis: association with malignancy. J. Rheumatol. 23, 101–105 (1996). 84. Hill, C. L. et al. Frequency of specific cancer types in dermatomyositis and polymyositis: a population-based study. Lancet 357, 96–100 (2001). 85. Ang, P., Sugeng, M. W. & Chua, S. H. Classical and amyopathic dermatomyositis seen at the National Skin Centre of Singapore: a 3‑year

86.

87.

88.

89.

90.

91.

92.

93.

94.

retrospective review of their clinical characteristics and association with malignancy. Ann. Acad. Med. Singapore 29, 219–223 (2000). András, C. et al. Dermatomyositis and polymyositis associated with malignancy: a 21-year retrospective study. J. Rheumatol. 35, 438–444 (2008). Chinoy, H., Fertig, N., Oddis, C. V., Ollier, W. E. & Cooper, R. G. The diagnostic utility of myositis autoantibody testing for predicting the risk of cancer-associated myositis. Ann. Rheum. Dis. 66, 1345–1349 (2007). Casciola-Rosen, L. et al. Enhanced autoantigen expression in regenerating muscle cells in idiopathic inflammatory myopathy. J. Exp. Med. 201, 591–601 (2005). Targoff, I. N. et al. A novel autoantibody to a 155‑kd protein is associated with dermatomyositis. Arthritis Rheum. 54, 3682–3689 (2006). Zampieri, S. et al. Polymyositis, dermatomyositis and malignancy: a further intriguing link. Autoimmun. Rev. 9, 449–453 (2010). Fujimoto, M. et al. Myositis-specific anti‑155/140 autoantibodies target transcription intermediary factor 1 family proteins. Arthritis Rheum. 64, 513–522 (2012). Trallero-Araguás, E. et al. Usefulness of anti‑p155 autoantibody for diagnosing cancerassociated dermatomyositis: a systematic review and meta-analysis. Arthritis Rheum. 64, 523–532 (2012). Allton, K. et al. Trim24 targets endogenous p53 for degradation. Proc. Natl Acad. Sci. USA 106, 11612–11616 (2009). Labrador-Horrillo, M. et al. Anti-TIF1γ antibodies (anti‑p155) in adult patients with dermatomyositis: comparison of different diagnostic assays. Ann. Rheum. Dis. 71, 993–996 (2012).

NATURE REVIEWS | RHEUMATOLOGY

95. Ichimura, Y. et al. Anti-NXP2 autoantibodies in adult patients with idiopathic inflammatory myopathies: possible association with malignancy. Ann. Rheum. Dis. 71, 710–713 (2012). 96. McCance, R. A. Osteomalacia with Looser’s nodes (Milkman’s syndrome) due to a raised resistance to vitamin D acquired about the age of 15 years. Q. J. Med. 16, 33–46 (1947). 97. Chong, W. H., Molinolo, A. A. Chen, C. C. & Collins, M. T. Tumor-induced osteomalacia. Endocr. Relat. Cancer 18, R53–R77 (2011). 98. Prader, A., Illig, R., Uehlinger, E. & Stalder, G. Rickets following bone tumor [German]. Helv. Paediatr. Acta 14, 554–565 (1959). 99. White, K. E. et al. The autosomal dominant hypophosphatemic rickets (ADHR) gene is a secreted polypeptide overexpressed by tumors that cause phosphate wasting. J. Clin. Endocrinol. Metab. 86, 497–500 (2001). 100. Folpe, A. L. et al. Most osteomalacia-associated mesenchymal tumors are a single histopathologic entity: an analysis of 32 cases and a comprehensive review of the literature. Am. J. Surg. Pathol. 28, 1–30 (2004). 101. Jiang, Y. et al. Tumor-induced osteomalacia: an important cause of adult-onset hypophosphatemic osteomalacia in China: Report of 39 cases and review of the literature. J. Bone Miner. Res. 27, 1967–1975 (2012). 102. Hill, A. B. The environment and disease: association or causation? Proc. R. Soc. Med. 58, 293–300 (1965). Author contributions B.M. researched the data for the article. B.M. and G.S. contributed substantially to discussions of the article content, writing the article and reviewing and/or editing the manuscript before submission.

ADVANCE ONLINE PUBLICATION  |  9 © 2014 Macmillan Publishers Limited. All rights reserved