Musculoskeletal disease in patients with severe acne has been described as early as 1961 (1). In 1972, an entity defined as chronic recurrent multifocal osteomyelitis (CRMO) was described by Giedion et al (2). Shortly thereafter, the association of CRMO with palmoplantar pustulosis (PPP) was noted (3). Following this description, wide arrays of cases were reported in the literature involving musculoskeletal symptoms associated with dermatologic disease, with >50 different names applied to these conditions. In 1987, Chamot et al attempted to unify the various descriptions of osteoarticular disease associated with skin manifestations into a syndrome with the acronym SAPHO: synovitis, acne, pustulosis, hyperostosis, and osteitis (4).
Since the description of SAPHO, controversy has existed as to whether the categorization of SAPHO as a distinct syndrome is appropriate. Some authors have argued that SAPHO is not a unique entity, but rather one component of a wide spectrum of disease (5). The basis of this argument is that there are overlapping clinical, radiologic, and pathologic characteristics that SAPHO shares with well-defined rheumatologic and dermatologic conditions, such as psoriatic arthritis and ankylosing spondylitis.
The purpose of this article is to critically weigh the evidence for SAPHO as a distinct entity, and to consider whether it should be properly considered a part of a larger spectrum of disease. To this end, we first review the clinical, radiologic, and pathologic features of SAPHO as described by investigators in this field. We then discuss current theories about the classification of SAPHO as a disease entity and attempt a synthesis of these approaches, particularly in light of the responsiveness of individuals with SAPHO to biologic therapies. Finally, we will explore future research directions, particularly with respect to how these future investigations may lead to a more definitive classification of SAPHO that is based on improved understanding of pathogenesis.
SAPHO: Cardinal Clinical, Radiologic, and Pathologic Features
The clinical features of SAPHO can be appropriately summarized by its descriptive acronym. To further develop these features, we will approach them individually, with the realization that SAPHO as a syndrome consists of a constellation of clinical features, each of which alone may not constitute a distinct clinical-pathologic disease.
The classic skin lesions seen in persons with SAPHO include PPP and acne (6). Acne is usually severe, and can present as acne conglobata, acne fulminans, or hidradenitis suppurativa. Skin manifestations may be discordant with musculoskeletal symptoms; there have been reports of lesions appearing up to 20 years after the arthritic manifestations (6, 7). Alternatively, the lesions may appear prior to the osteoarticular manifestations. In other cases, outbreaks of skin lesions seem to coincide with flares of arthritic symptoms. Pustular psoriasis has also been described as a manifestation of SAPHO (6). This can be difficult to differentiate clinically or histopathologically from PPP. Psoriasis vulgaris has also been seen in association with the hyperostotic and osteitic features of SAPHO (6). The possible interrelationship of psoriatic arthritis (PsA) and SAPHO will be discussed below.
In addition to synovitis, hyperostosis and osteitis are seen in SAPHO, as implied by the acronym. These latter conditions are the result of a chronic inflammatory reaction involving the medulla and cortex of the bone (7). Osteitis is believed to be the cornerstone of diagnosis in SAPHO. Clinically, the first clues suggesting osteitis arise when bone pain is discriminated from joint pain on the basis of a careful physical examination. Endosteal and periosteal thickening characterize the bony lesions. There is diffuse cortical thickening and narrowing of the medullary canal, and areas of osteolysis may be present. In addition to the bony manifestations of SAPHO, inflammatory enthesopathy has also been reported as a feature (6).
Certain osteoarticular sites appear to be predisposed to involvement in SAPHO. Axial arthritis is common, reported in up to 91% of cases (8). The anterior chest wall is a classic location of involvement, in particular, the clavicles, sternum, and sternocostoclavicular joints. In terms of peripheral arthritis, involvement of the knees, hips, and ankles commonly occurs (8). In addition, spondylodiscitis has been described (8). Mandibular and pelvic involvement are also prominent, and sacroiliitis, often unilateral, can be seen (8).
The typical presentation in the pediatric population is CRMO (8). The differentiating clinical features of pediatric CRMO and adult SAPHO are mainly in localization of inflammation: in pediatric CRMO, this inflammation is seen more often in the extremities than in the axial skeleton, with rare involvement of the jaw and sternum (9). Multiple or symmetric lesions of the long bones have also been described (8). One of the key issues to be resolved in future studies is a comparative analysis of the natural history of pediatric CRMO and adult SAPHO, particularly a comparison of long-term outcomes. Indeed, SAPHO and CRMO were considered to be distinct entities for many years. However, most clinicians now agree that the manifestations of CRMO and SAPHO justify inclusion in the same nosologic group (10, 11).
The early bone lesions seen in SAPHO are histologically indistinguishable from those seen in osteomyelitis (6), and this is usually the key differential diagnosis and frequently occasions the need for a biopsy. Later in the course, there is a characteristic infiltrate of mononuclear cells. In the late phase, sclerotic, enlarged trabeculae are seen along with increased marrow fibrosis (6).
In terms of skin lesions, neutrophilic pseudoabscess is a consistent, connecting feature of the various dermatologic manifestations of SAPHO (6). Whether there are histopathologic features that confidently discriminate these lesions from pustular psoriasis has remained controversial among pathologists.
A wide variety of radiographic findings have been described in SAPHO. These include anterior chest wall lesions (65–90% of patients), most frequently in the sternocostoclavicular region (8). On scintigraphy, the presence of a bull's head–shaped enhancement is proposed to be highly specific for SAPHO (8). Several spinal lesions, such as spondylodiscitis, osteosclerosis of vertebral bodies, and paravertebral ossifications, have been described. Sacroiliac joints are seen to be involved, often in a unilateral pattern, with inflammatory changes in 13–52% of cases (8). In 30% of cases there is long bone involvement, predominantly affecting the metadiaphyseal region of the distal femur and proximal tibia (8). Periostitis is a characteristic radiologic feature, with images that can be indistinguishable from osteomyelitis. Flat bones, such as the ilium and mandible, also can be involved, displaying diffuse sclerosis. Peripheral arthritis is seen, but joint destruction is rare. Knees, hips, and ankles are most frequently involved. Early radiographic changes include juxtaarticular osteoporosis; later, joint space narrowing and marginal or central erosions may be seen (8). Ankylosis, hyperostosis, and enthesopathy may follow.
Possible link to an infectious agent
Numerous investigators have brought forth the theory that the osteitis and dermatitis seen in SAPHO are the results of a persisting pathogen of low virulence, or that the syndrome is triggered by a pathogen and sustained by an autoimmune response subsequent to that infectious challenge (12). In some studies, Propionibacterium acnes has been isolated from synovial tissue or fluid and bone biopsy specimens, suggesting a possible etiologic role for this organism (13). However, because P acnes is commonly recovered from skin cultures, it is possible that these findings reflect contamination of a sample that has been obtained percutaneously. Generally, antibiotic therapy has not been subjected to randomized controlled trials. There are uncontrolled reports of sustained response to doxycycline (14) and azithromycin (15).
New information about the pathogenesis and classification of SAPHO is accumulating from advances in the genetics of the syndrome and related diseases. In 1991, a murine model was developed that displayed the characteristic chronic multifocal aseptic osteomyelitis seen in SAPHO syndrome (16). The mutation in this model has subsequently been mapped to chromosome 18 and found to affect proline-serine-threonine phosphatase interacting protein 2 (PSTPIP2) (17). In a similar vein, a susceptibility gene from CRMO has been localized to chromosome 18q21.3-18q22 (18).
Light may also be shed onto the pathogenesis of SAPHO by examining the genetic principles of diseases that are similar in nature. For example, Majeed syndrome, a genetic human disease first described in 1989, is characterized by neutrophilic dermatosis, multiple osteitic lesions, and abnormalities of erythropoiesis (19). Subsequent research has localized a potential genetic mutation that predisposes to Majeed syndrome on LPIN2, a gene that encodes lipin 2 (20). Lipin 2 may be involved in the apoptosis of polymorphonuclear cells (20).
Similarly, pyogenic sterile arthritis, pyoderma gangrenosum and acne (PAPA) syndrome has characteristics that may lend insight into the pathogenesis of SAPHO. PAPA has been found to be transmitted in an autosomal dominant fashion (21), and the predisposing mutation is located on chromosome 15 (22). This mutation affects CD2-binding protein/PSTPIP1 (23).
Future studies into the genetics of SAPHO, particularly in comparison with these other diseases and murine models, may shed important light on the debate about the proper classification of SAPHO among the rheumatic diseases.
The Classification of SAPHO as a Distinct Disease Entity
One central question that has been debated in the case of the SAPHO syndrome is where it fits in a classification of rheumatic diseases. Is it an entity in itself? If so, does it constitute a subset within the family of spondylarthropathies (SpA), or is it best considered a variant of another rheumatic disease such as PsA? This section will review theories in this regard.
A proposed classification system for the spondylarthropathies is shown in Figure 1. Within this family are the well-characterized subsets ankylosing spondylitis (AS), PsA, reactive arthritis (ReA), and enteropathic arthritis (EA). There is a significant subset of patients with SpA who do not fulfill the diagnostic criteria for these clinical entities and who are defined as having undifferentiated SpA. Using this classification system as a guideline, one can consider several different options for correctly placing SAPHO in the spectrum of seronegative arthritis: 1) SAPHO is a subset of SpA that overlaps AS and PsA, 2) SAPHO is a subset of PsA, 3) SAPHO is an SpA that overlaps ReA, 4) SAPHO is a distinct and unique subset of SpA, and 5) SAPHO is not an SpA.
SAPHO is a subset of SpA that overlaps PsA and AS
There are aspects of SAPHO that are common in AS, in particular, the features of ankylosis and enthesitis that are characteristic of AS. The axial skeleton is most frequently involved. Furthermore, radiologic findings often show sacroiliitis as in typical AS (13–52% of cases) (8). Paravertebral ossifications are seen in SAPHO, as in AS. In some cases, the skin osteoarticular manifestations precede the skin features by years, making the clinical picture more suggestive of AS than of the SAPHO syndrome. In some studies, HLA–B27 has been reported to be positive in 13–30% of SAPHO patients (6, 12); in other studies, there was no difference in HLA status between patients with SAPHO and controls (6). Because HLA–B27 is present in 90% of patients with AS, this genetic marker would argue for separating SAPHO from AS.
With respect to PsA, there are many overlapping features with SAPHO. For example, the skin manifestation in PPP is histologically identical to that of pustular psoriasis. However, the radiographic appearance of osteitis and hyperostosis is not usually seen in PsA. To further support the possible link between SAPHO and PsA, a French multicenter study demonstrated that although psoriasis was not a prominent feature in patients classified as having SAPHO, there was 3 times the amount of psoriasis among those patients compared with the general population (6). This finding may imply a connection between the genetics of psoriasis and the genetics of the SAPHO syndrome.
SAPHO is a subset of SpA that is encompassed by PsA.
With PPP being so comparable with pustular psoriasis, it could be argued that the arthritic manifestations related to PPP actually represent PsA. Several investigators have reported psoriasis vulgaris with the bone lesions seen in SAPHO (6). In some cases, psoriasis vulgaris has developed after the initial typical skin changes and osteoarthritic manifestations. However, the classification of SAPHO as a form of PsA does not take into account patients with acne as the primary skin manifestation. Conversely, the distinctive nail dystrophy of PsA has not been reported in SAPHO.
SAPHO is subset of SpA that overlaps with ReA.
The possible link to infection with an organism of low virulence, such as P acnes, lends credence to this theory. Also, typical changes of osteomyelitis are seen even in cases where a causal organism is not isolated. Whether the arthritic changes seen in SAPHO could be attributable to infection or autoimmune reaction to pathogenic material remains an unanswered question.
SAPHO is a distinct and unique SpA.
This position champions the idea that although SAPHO may be related to the other spondylarthropathies, it is an entity of its own. While similarities exist between other SpA conditions, there are enough unique features in SAPHO to classify it as its own disorder (6). In support of this position, the distinctive presentations of SAPHO are sternoclavicular hyperostosis, arthro-osteitis related to PPP, and severe acne. There is a shared basic pathologic process in both PPP and acne, namely, neutrophilic pseudoabscess (6). In patients with PPP, acne, or even no skin involvement, the bone involvement has a distinct and consistent pattern, both radiologically and anatomically. There is preferential localization to the anterior chest wall. Furthermore, pseudoseptic arthritis can be observed with acne, PPP, and psoriasis. Sacroiliac joint involvement has been reported in the setting of all 3 skin conditions (in 13–52% of cases) (8).
SAPHO is not an SpA.
This last position states that SAPHO is not an SpA at all, but an entity outside of that group. With similarities between SAPHO and the various subsets of SpA as described above, this position is perhaps the least tenable of those proposed above. However, the following should be noted about SAPHO: there is no HLA–B27 association; there is no associated acute anterior uveitis; there is no strong familial association; sacroiliitis is an infrequent manifestation, and syndesmophytes and spinal ankylosis are rare; and there is no male predominance. These features make SAPHO somewhat of a poor fit into the SpA family.
SAPHO: A Spectrum?
It is clear that SAPHO is an entity that has many features that fit into a variety of already established disease categories. We submit that it is possible that conceptualizing the different types of SpA as separate disorders may be a flawed model. It may be more accurate instead to regard these conditions as lying along a spectrum of disease (Figure 2). In this model, there are not discrete categories of SpA, but rather a spectrum or gradation of symptoms that encompasses many features seen in varying conditions. SAPHO would therefore fit in the area of overlap between several of these conditions. As described above, features of SAPHO overlap with many different SpA. If SAPHO is viewed as part of a spectrum of disease, various clinical presentations may be explained by their alignment along this gradient. This model is appealing because it may be more accurate for describing SAPHO than those proposed earlier in this article. It is important to note, however, that this model describes the personal point of view of the authors and has not been validated.
Indirect support for the spectrum model of SpA derives from the similar response of these apparently different entities to anti–tumor necrosis factor (anti-TNF) therapies. Undifferentiated SpA, AS, PsA, ReA, and EA have all been shown to be responsive to anti-TNF agents. Similarly, case reports of the response of patients with SAPHO to these same medications are now appearing (24, 25). The fact that blocking the same factor in the inflammatory cascade provides similar clinical responses in these clinically diverse entities provides indirect support for some commonality underlying these conditions. It may be, however, that blocking TNF represents an interruption in nonspecific inflammatory pathways, regardless of whether diseases are pathophysiologically similar or not.
In the evolving nomenclature of rheumatic diseases, it will increasingly be seen that common pathophysiologic mechanisms will be unifying and discriminating principles. Many inflammatory conditions traditionally thought to be fundamentally different are proving to share an important clinical aspect, namely, responsiveness to anti-TNF agents. However, this is a story still evolving. Among these rheumatic diseases, there remains a substantial proportion of patients who respond suboptimally to anti-TNF therapy, and even among responders these agents are remittive but not curative. Furthermore, whether the natural history of the disease long term is fundamentally changed has yet to be defined.
Thus the challenge still remains for pathogeneticists to pursue the basic genetic and molecular basis of challenging syndromes such as SAPHO. In the process of defining these fundamental mechanisms, current classification schemes based on commonality of clinical features will be replaced with those based on the underlying disease process. Even with our current limited information, there are enough tantalizing possibilities in exploring the possible classification of SAPHO to make it a worthy target of future study, both on a clinical and on a biologic basis.