The initial literature research revealed a total of 925 articles. All 829 articles written in English or German were subsequently screened, which resulted in 81 articles fulfilling the inclusion criteria. After a detailed review and application of the previously determined exclusion criteria, 22 articles were considered suitable for analysis. Figure 1 provides an overview of the process of literature research and selection.
All included studies were published between the year 2000 and 2020 and consisted of 8 case series, 6 case reports, 4 cohort studies, 2 case control series, 1 interventional study and 1 randomized controlled trial with < 80% follow-up. The level of evidence ranged from level 5 to level 2b.
Patient characteristics
The included studies comprised 521 patients with a mean age of 61.2 years, of whom 332 (63.7%) were female. The medial femoral condyle was affected in 70.6% of cases. The majority of lesions were in Koshino stage 1 (67.6%) or 2 (27.5%) whereas only 4.9% of cases were classified as stages 3 and 4. The average duration of therapy and follow-up time were 4.3 months and 3.0 years, respectively. With an average of 7.2 months, drug therapies were used longer compared to other therapies (mean of 2.9 months). Detailed descriptions are shown in Tables 1-5.
Therapies
The following therapy concepts were used: RWB or even complete unloading by using crutches, bed rest, specifically designed shoes or shoe insoles, unloader braces, etc. (n=11), bisphosphonates (alendronate/ibandronate) (n=3), a combination of RWB (crutches/cane/walker) with the administration of bisphosphonates (alendronate/ibandronate/zoledronate) (n=4), teriparatide (n=1), laser-needle therapy (n=1), pulsed electromagnetic fields (PEMF) therapy (n=1) as well as a hyperbaric chamber combined with orally applied ibandronate and RWB (n=1). In some studies, these therapies were complemented by secondary assistive treatment approaches, which included the administration of NSAID, vitamin D and/or calcium, injections of hyaluronan or corticosteroids and physiotherapy.
Magnetic resonance imaging
The MRI-based criteria used for the differential diagnosis of SONK were found to be heterogeneous. While some authors screened images for specific features, such as focal hypointense areas on T1- or T2-weighted images, visible fracture lines or focal epiphyseal contour depressions [7,13–24], others diagnosed SONK in the presence of bone marrow edema (BME) accompanying typical symptoms [25–27]. Two case reports radiologically demonstrated the primary overlay of fracture lines by BME [28,29]. Some studies failed to report the MRI criteria used for diagnosis [30–33]. In total, BME were observed in 92.9% of patients, whereas fractures were reported to be present in 44.7%.
Prognostic/risk factors
In the included studies, the following parameters were found to negatively influence the course of the disease: age over 65 years [24], varus alignment [17,19,24], affection of the medial femoral condyle or medial tibial plateau [24], meniscal extrusion [24], end stage osteoarthritis [24] and larger defect size [18,19]. Decreased bone mineral density [18,19] was associated with the onset of SONK, but not with its progression [17–19]. Obesity [17,24], sex [17,24] or the presence of BME were not associated with poorer prognosis. Contradictory findings were reported for age [17], meniscal extrusion [19] and lesion size [7,17,27,31].
Therapy outcomes
Sixteen of the twenty-two included studies (73%) reported positive clinical and/or radiological outcomes [7,13,17,20–23,25–33]. Although RWB was the most thoroughly tested treatment modality, it has also been challenged by some authors. While four studies reported only moderate to poor results [14–16,24], nine observed favorable outcomes when testing one of the various forms of RWB [7,17,21,27,32] or a combination of RWB with bisphosphonates [20,25,29,33]. Treatment through bisphosphonates alone was positively evaluated in all studies [23,30,31]. The only controlled trial, however, found bisphosphonates not to yield superior outcomes as compared to placebos [23]. Furthermore, even though orally administered alendronate was the most often used bisphosphonate, the available data and the complexity of the individual therapies did neither allow to establish a superior bisphosphonate, nor the best mode of administration or treatment regimen. The administration of teriparatide [22], laser-needle therapy [13], PEMF therapy [26] and the usage of a hyperbaric chamber [28] were all successfully tested in single trials only.
Regarding different MRI differential diagnoses, it was found that especially lesions showing only BME responded excellently to conservative therapy measures, whereas treatment success was less clear in the presence of further MRI features (Table 6). The only frequently described MRI feature for which mainly moderate to poor results were reported, are focal low signal intensities on T1-weighted images [14,15,19,24]. Although a complete remission of pathologic MRI signals was reported 3 months after therapy initiation [7], most lesions persisted longer than 6 months despite radiologic signs of healing [7,13,22,25,26,29–31]. In this regard, Geijer et al. [29] reported the course of a lesion the healing of which took over 2.5 years. By contrast, clinical symptoms often improved significantly after 2 to 6 weeks [13,21,22,25,29,30,33]. A complete disappearance of symptoms was described at the earliest after 2 to 3 months of continuous therapy [7,13,17]. Longitudinal data, allowing for the development of the stage of SONK to be traced, were available for 88 knees. Out of the 82 lesions initially diagnosed at stage 1, 19 progressed to more advanced stages. While ten of these lesions progressed to stage 4 and required knee arthroplasty, nine knees had good clinical results. However, due to marked symptomatology or radiological progression, a total of 158 out of 521 patients (30.3%) required surgery during follow-up.