Literature search

The existing literature mostly comprised observational studies reporting a diverse set of treatments considering variable (mostly descriptive) outcomes.6 52 Evidence from randomised controlled trials (RCTs) was limited to a Chinese study, which had to be excluded from this review due to publication language (Chinese).53

An initial search delivered 2290 records, of which 1123 manuscripts remained after removing duplicates and non-English publications that underwent title and abstract screening. Of these, 1080 manuscripts were excluded applying exclusion criteria. Remaining manuscripts (N=43) were accessed for a full-text screen. Thirteen publications were removed as they were conference abstracts without an associated full text manuscript (remaining N=30), and two additional publications were removed based on exclusion criteria. Finally, 28 studies were included in full-text review (figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

PRISMA diagram displaying search results, exclusion and inclusion of manuscripts.45 PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Demographic information and reported outcomes

The 28 studies selected for full review included 796 patients, aged 5–84 years. Twenty-five papers detailed sex distribution (including n=553/796, 69.5% of patients) with approximately 2.2:1 female predominance (female n=374/553, 67.6%; male n=179/553, 32.4%) (online supplemental table S2).

Sixteen manuscripts originated from Europe, eight from China and two from Australia. Remaining studies were conducted in Tunisia (N=1) and Israel (N=1). Nine papers detailed ethnicity (including n=444/796, 55.8% of patients). Of these, the majority were White Caucasians (n=223/444, 50.2%), followed by Asian ethnicity (n=202/444, 45.5%, of whom 188/202, 93.1% were Chinese). A minority of reported patients were of black ethnicity (n=19/444, 4.3%).

Outcome measures applied across 28 studies were heterogeneous and varied in quality and quantity (online supplemental table S3,S4). Most studies focused on outcomes relating to osteoarticular manifestations, such as pain improvement via Visual Analogue Scale (VAS) pain scores (10/28 studies, 35.7%), ‘other’ patient-reported outcomes (8/28, 28.6%) and need for treatment escalation (4/28, 14.3%). Global disease activity was most commonly assessed through the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI, 7/28, 25%), Ankylosing Spondylitis Disease Activity Score (ASDAS, 2/28, 7.14%) and number of flares after treatment initiation (3/28, 10.7%). Other measures included patient questionnaires for osteitis-associated complaints, Physician Global Assessment, Health Assessment Score (HAS), and physician assessed osteitis and skin disease activity. Evidence of osteoarticular involvement on MRI was the most commonly reported imaging outcome (6/28, 21.4%), followed by changes on plain X-rays (3/28, 10.7%), and CT including cone beam CT (CBCT, an imaging modality developed specifically for imaging teeth, jaws and temporomandibular joint) (3/28, 10.7%).23 Laboratory inflammatory markers reported usually included Erythrocyte Sedimentation Rate (ESR) (10/28, 35.7%) and C reactive protein (CRP) (9/28, 32.1%). Fewer studies addressed skin and nail manifestations (10/28, 35.7%). Most of these studies reported descriptive measures (6/10, 60%), however, Nail Psoriasis Severity Index (NAPSI, 2/10, 20%) and Palmoplantar Pustulosis Area and Severity Index (PPPASI, 3/10, 30%) were also used.

Treatments

Several treatments were reported across the 28 studies either alone or, most frequently, in combination.4–6 23 26 Likely because of variable involvement of various organ systems (predominantly skin and bone), symptomatic improvement was frequently associated with using multiple drug therapies in parallel. Bisphosphonate use was reported in 17/28 (60.7%) studies (pamidronate 12/17; alendronate and zoledronic acid 1/17; not specified 4/17).2 3 5–7 12 13 15 16 19–21 23 26 29 30 54 NSAIDs (11/28, 39.3%), cDMARDs (10/28, 35.7%), including methotrexate, sulfasalazine and azathioprine, and bDMARDs, including TNF inhibitors (TNFi, 11/28, 39.3%) and IL-17 blockers (IL-17i, 2/28, 7.1%%), were discussed. One study reported the use of Janus kinase (JAK) inhibition with tofacitinib.18 No serious adverse events were reported for treatments used (online supplemental table S5).

Antimicrobial agents

Historically, infections (especially with Propionibacterium acnes) had been considered as a pathomechanistic feature of SAPHO.8 52 As a result, 85 patients from 6 studies were treated with antimicrobial agents none of which resulted in sustained remission4 12 19 25 26 28 (figure 1). Azithromycin and doxycycline were used in several studies despite negative bacteriological bone cultures, likely to address skin colonisation associated with acne. Unsurprisingly, only a minority of patients (11/55, 20%) reported (time-limited) symptomatic improvement.4 12 19 26 28 In patients with positive bacterial cultures for P. acnes in bone biopsies, antimicrobial treatment improved osteitis lesions and reduced skin disease activity after 16 weeks. However, disease activity returned to baseline following discontinuation of antimicrobial treatment.25 In one study, despite negative or unknown bacterial cultures for P. acnes, antimicrobial therapy with clindamycin in combination with NSAIDs (below) resulted in clinical improvement with persistent pain resolution (4/4, 100%) and improved imaging findings (osteosclerosis n=2; osteitis n=2; osteolysis n=1) after 3 months or 4 months treatment. Another patient had resolution of sternoclavicular joint swelling and tenderness after 8 months of treatment.4 However, it remained unclear whether effects were due to antimicrobial agents and/or NSAIDs.

Only few and mild adverse effects were reported, including not further specified ‘intolerance’ (doxycycline n=2; azithromycin n=1).12 25

Non-steroidal anti-inflammatory drugs

NSAIDs are frequently used in rheumatology to modulate pain, inflammation, and osteoclast activity through inhibition of prostaglandin synthesis41 55 56 (figure 1). Across studies, NSAIDs were usually used as first-line treatment to provide pain relief in SAPHO.6 8 52 A small proportion of SAPHO patients achieved partial or full remission in response to NSAID treatment alone, including 20% (4/20) of patients in a small Belgian cohort.2 5 12 19 26 29 30 The majority of studies, however, reported inadequate effects or incomplete improvement.5 21 30 Of those studies, reporting efficacy of NSAIDs, most focused on pain reduction.2 4 6 8 28 One study also reported improved radiological and dermatological features (in addition to pain resolution) in patients treated with a combination of the antimicrobial agent clindamycin and lornoxicam.4

Side effects of NSAIDs included gastrointestinal symptoms, but no severe toxicity.21 26

Corticosteroids

Corticosteroids inhibit phospholipase A, which results in reduced prostaglandin E production affecting pain, inflammation and osteoclast function. Furthermore, they alter the expression of NFκB-dependent proinflammatory cytokines and enhance anti-inflammatory cytokine expression (including IL-10)41 56 (figure 1). Long-term use of corticosteroids is limited by side effects that, among others, include loss of bone density, thinning of skin, weight gain and the development of iatrogenic Cushing’s syndrome, infections, etc.57 Corticosteroids can be applied systemically or locally (topical use on the skin, joint injections).52

Across 7 studies, a total of 56 SAPHO patients were treated with corticosteroids. Topical corticosteroids were used for dermatological manifestations, with psoriasis vulgaris resolution in one case when taken alongside adalimumab and eosin.5 Intra-articular corticosteroid injections were reported by four studies. Three studies suggested intra-articular corticosteroids to be effective, providing greater osteoarticular improvement when compared with oral therapy.12 17 26 One study failed to identify positive effects on osteitis (p=0.06) using 20 mg intra-articular triamcinolone but suggested benefit for arthritis management.17

Systemic corticosteroids were used with transient and/or partial improvement in 26 individuals from three studies.2 26 29 Favourable outcomes were reported in individuals when corticosteroids were given in combination with either adalimumab (psoriasis vulgaris resolution, 1/1), Tripterygium wilfordii Hook F (a traditional Chinese herb with immunosuppressive effects: muscle oedema regression, 4/4; subcutaneous soft tissue swelling regression, 1/1; bone marrow oedema (BMO) regression, 2/4; periosteitis regression, 1/2; temporomandibular joint synovial thickening regression, 3/3) or methotrexate (cortical/medullary lysis on CBCT, 1/1).5 23 In a small Chinese cohort, corticosteroid monotherapy led to regression of muscle and BMO (both 2/3, 66.7%), subcutaneous soft tissue swelling (1/1) and temporomandibular arthritis (1/1).23

While no adverse events or severe side effects were reported in these included studies, long-term follow-up is lacking, and concerns remain in relation to drug safety and recognised corticosteroid-associated complications.58

Bisphosphonates

Bisphosphonates inhibit osteoclast-mediated bone resorption and have anti-inflammatory effects, reconstituting the balance between proinflammatory and anti-inflammatory cytokines.3 6 7 13 15 52 A total of 176 patients were treated with bisphosphonates across 17 studies. Notably, reported effects of bisphosphonates are mostly limited to bone inflammation with no consistent improvement of cutaneous lesions across the literature.59 60

Treatment with bisphosphonates improved osteoarticular disease in a majority of patients across studies.2 3 6 7 12 13 15 16 19–21 23 26 29 30 This included patients refractory to NSAIDs and corticosteroid injections.12 13 26 Active agents, administration methods and doses vary between studies (table 2). Four manuscripts did not disclose specific bisphosphonates used, route of administration, nor dose.23 26 29 30 Improvement was described as partial or complete pain resolution.12 13 26 29 30 Where available, MRI confirmed rapid and sustained improvement of spinal BMO in two studies including 44 patients (for 37 of whom MRI studies were available).7 23 Somewhat unexpectedly, one open-label study also reported beneficial effects of pamidronate on skin lesions, preventing recurrence of pustulosis in 9/10 patients (90%) refractory to oral corticosteroids, colchicine, methotrexate, sulfasalazine or infliximab.13 However, NSAID use was permitted as concomitant treatment, which may have had additional effects.13 Others did not confirm effects on the skin.15

Across studies, pamidronate was the most frequently chosen bisphosphonate (n=122/176 patients, 69.3%; 12/17 studies, 70.6%). Pamidronate was effective in early disease, but may also have a role in late-stage otherwise treatment refractory SAPHO.15 16 Intravenous pamidronate effectively controlled osteoarticular lesions, with most patients experiencing partial or complete pain control.13 15 16 19 20 Some studies reported reduction of osteolytic lesions coinciding with complete clinical, laboratory and radiological remission.2 23 In one study, pain relief was associated with reduced need for other medications, including NSAIDs.16 A study in paediatric SAPHO patients suggested persistent effects of pamidronate with improved activity and well-being.3

Though not included across all international treatment plans for paediatric CNO/CRMO because of lacking information on anti-inflammatory effects,61 62 alendronate and zoledronic acid have been used in SAPHO in a Chinese study and an Australian study.6 54 Symptomatic relief has been reported in 16/18 (88.9%) patients in the Chinese study (70 mg alendronate per week (15 recipients) or 5 mg zoledronate annually (3 recipients). However, the benefit from bisphosphonate alone is difficult to ascertain as 17/18 (94.4%) received concomitant therapy with NSAIDs, DMARDs, low-dose corticosteroids or TNFi.6 Alendronate monotherapy was effective in one Australian patient.6 Three patients from the same study received either pamidronate or zoledronic acid, which did not improve bone involvement.54

Bisphosphonate associated side effects were in line with previous reports and included influenza-like symptoms and/or timely limited feverish episodes,3 5 7 12 16 headaches and mild hypocalcaemia7 in patients receiving pamidronate. Gastrointestinal disturbances were infrequently reported in Chinese cohorts. However, one patient received cDMARDs alongside alendronate.6 7 Notably, one study from Italy reported no side effects in a cohort of 14 patients.15 Taken together, bisphosphonates appear safe and promising, particularly with spinal BMO.7 23 54

Conventional (non-biologic) DMARDs

cDMARDs are usually used when symptoms persist despite NSAID treatment.52 Across 10 studies, 173 patients were treated with variable effects. Methotrexate and sulfasalazine reduce inflammatory cytokine expression, thus potentially correcting the imbalance between proinflammatory and anti-inflammatory signals.63 Methotrexate and sulfasalazine were most commonly used.6 12 19 23 26–30 54 Two studies reported colchicine use (in 34 patients) with little or no improvement.12 26 Azathioprine, leflunomide and cyclosporine were reported in a few individual patients.6 29 30 54

Three studies suggested beneficial effects of methotrexate on osteoarticular and cutaneous involvement.19 26 54 Other studies suggested that using NSAIDs or corticosteroids alongside methotrexate may be more effective than methotrexate alone.6 23 29 Overall, complete inefficacy of methotrexate was relatively uncommon (11/54 patients from 6 studies, 20.4%).6 12 19 26 29 54 Among these, 8/13 patients relapsed following good initial response,54 and 2 switched to other cDMARDs without additional benefit.29

Sulfasalazine was associated with minimal or partial improvement of pain, palmoplantar pustulosis and acne.12 26 Some patients refractory to NSAIDs experienced resolution of osteoarticular pain.19 One patient had over 17 years sustained benefit as demonstrated by Physician Global Assessment scores.54

Overall, cDMARDs were tolerated without severe side effects. Most publications did not report side effects beyond gastric discomfort in patients receiving methotrexate or sulfasalazine.6 One study suggested that methotrexate (without concomitant corticosteroids) may be associated with metabolic syndrome.28

Biological DMARDs

bDMARDs are used in otherwise treatment refractory patients. Cytokine blocking strategies are used to (at least partially) correct the cytokine imbalance associated with SAPHO (figure 1). 2 5 6 12 14 19 23 24 27 29 30 54 64

A total of 90 patients from 11 studies were treated with TNFi. Overall, reported effects were positive.6 14 19 27 54 In one case where adalimumab lost efficacy with worsening skeletal and cutaneous symptoms after initial pain and palmoplantar pustulosis improvement, add-on cDMARD therapy with methotrexate, sulfasalazine and eventually cyclosporine led to symptom relief.6 Reports of inefficacy were few (n=5/90, 5.56%),6 12 14 29 and included an individual who had initial clinical response but ultimately had to add on cDMARD treatment6 and two patients who initially benefited from infliximab before switching to alternative TNFi for recurrence of bone pain.14

Most studies reported partial or complete clinical response, including reduction of pain, disease activity as measured with BASDAI and other osteoarticular presentations including sacroiliitis.2 5 6 14 30 54 64 Skin improvement was described in four Chinese individuals.64 The use of etanercept was associated with complete clinical and laboratory/imaging remission in most patients of a small Chinese cohort (7/10, 70%), but did not allow discontinuation of concomitant therapy in a French study.2 27 Improvement was sustained across several studies, including pain control following treatment discontinuation,64 and preventing joint and spinal damage progression (8/11; 73%) of patients continuing treatment for at least 1 year.54 Infliximab led to persistent physician-assessed benefit for 3 years in one individual.12 Adalimumab remained efficacious with reduced BASDAI for 22 months in another individual.14

Side effects were rare in patients treated with TNFi. Nine cases (9/90, 10%) of psoriasiform lesions were noted across two studies, including paradoxical psoriasis, existing palmoplantar pustulosis exacerbations and urticaria.14 64 Late onset of skin manifestations during the disease course were associated with increased risk of exacerbated lesions under TNFi in the Chinese study.64 Most paradoxical lesions resolved after discontinuing TNFi and introduction of non-biological therapy.14 64 However, at least some of these potential adverse events may have been associated with the natural course of disease in some patients as SAPHO is associated with psoriasiform skin disease.

Additional cytokine blocking agents targeting IL-6 (a factor during effector Th17 T-cell differentiation) or the IL-23/IL-17 axis have been considered for SAPHO because of clinical overlaps with psoriasis and spondylarthritis.23 24 54 Nine patients were treated across three studies from China, France and Australia. Blockade of the IL-6 receptor with tocilizumab failed to induce bone lysis remission on CT in a Chinese SAPHO patient.23 Outcomes in response to IL-17 blockade with secukinumab or ixekizumab were disappointing as none of the three Australian patients reached full and sustained remission.54 Ustekinumab targets IL-12/23 and led to cutaneous benefit in 33.3% (1/3) of a small French cohort but was not reported to benefit bone features; secukinumab had benefited 66.7% (2/3) who had not received ustekinumab in the same study.24 One case of paradoxical psoriasis (2/9 patients, 22.2%) was reported for each secukinumab and ustekinumab.24

Protein kinase inhibitors

JAKs are involved in cytokine receptor signalling, and the differentiation and activation of effector Th17 cells (figure 1). Because of clinical overlaps between SAPHO, psoriasis and spondyloarthropathies, they may be promising treatment options in patients refractory to other treatments.

One study investigated effects of the JAK inhibitor tofacitinib on dermatological manifestations in 13 Asian female SAPHO patients (12 weeks of 5 mg tofacitinib, twice daily).18 Results were positive with significant benefit for nail lesions, palmoplantar pustulosis and associated quality of life. While a reduction of systemic inflammatory parameters was observed (CRP and ESR), no information was provided on bone involvement.

One patient experienced acute tonsillitis during the treatment period. No severe adverse events were reported.

Other treatments

Tripterygium wilfordii Hook F is a Chinese herbal medicine used to treat immunological disorders.6 22 The mechanisms underlying its effects are diverse and remain largely unknown. It has been suggested to affect effector T-cell function, monocyte phenotypes, and proinflammatory and anti-inflammatory cytokine expression.65 Lacking regulatory approval, its potential toxicity and side effects limit clinical applications in the Western world. Three Chinese studies reported 41 SAPHO patients treated with Tripterygium wilfordii Hook F. Treatment associated with improved disease activity (measured by ASDAS and BASDAI) and reduced pain when compared with baseline across two small cohorts.22 23 Results were less convincing in the third study, including lack of response in one patient.6 Only one of the three studies reported side effects with elevated transaminases (ALT levels) in 14/30 patients (46.7%), of whom one individual had to stop treatment.22

In addition to systemic anti-inflammatory treatment, three studies reported patients receiving topical treatments for skin involvement, including psoralen in combination with ultraviolet light radiation (PUVA), retinoids (systemically or topically), betamethasone and calcipotriol (a vitamin D derivative containing ointment).4 26 30 Of the two studies reporting patient numbers, four patients were treated, however, treatment efficacy was only discussed in one patient reporting palmoplantar pustulosis improvement following treatment with topical corticosteroids and PUVA.4 Lastly, Xiang et al suggested a possible association between tonsillectomy and improved cutaneous and osteoarticular manifestations, as ostealgia and palmoplantar pustulosis significantly improved with reduced VAS and PPPASI scores in 6/7 and 6/6 patients, respectively.66