Objectives To determine the prevalence of bone marrow oedema (BME) at the sacroiliac joint (SIJ) in early postpartum (EPP), nulliparous (NP) and late postpartum (LPP) women, and to identify factors associated with BME presence at the SIJ.
Methods Three groups were obtained: NP (never given birth), EPP (given birth within 12 months) and LPP (given birth more than 24 months). The primary outcome was the presence of BME and/or structural lesions (erosions, osteophytes, ankylosis and sclerosis) at the SIJ MRI.
Results BME prevalence was greater among EPP (33%) than NP (14%, p=0.001), but was not different to LPP (21%, p=0.071). The Assessment of SpondyloArthritis international Society (ASAS) MRI criteria for sacroiliitis were positive in 75%, 71% and 80%, respectively, of EPP, NP and LPP women with BME. EPP (38%) had similar prevalence of sclerosis than LPP (28%, p=0.135), but greater than NP (18%, p=0.001). Lastly, EPP (28%) had similar prevalence of osteophytes than LPP (42%) and NP (27%), although there was a difference between LPP and NP (p=0.006).
Conclusions EPP have higher BME prevalence at the SIJ than NP, EPP tend to have higher BME prevalence compared with LPP and BME presence decreases with time from delivery. Three-quarters of women with BME at the SIJ had a positive ASAS MRI criteria for sacroiliitis, indicating that BME presence as the main criterion for a positive diagnosis can lead to false-positive results. SIJ MRIs should not be interpreted in isolation, since age, time from delivery and other factors may outweigh the pertinence of MRI findings.
Trial registration number
- magnetic resonance imaging
- low back pain
Statistics from Altmetric.com
What is already known about this subject?
Recent studies suggested that bone marrow oedema (BME) and degeneration are frequently observed at the sacroiliac joint (SIJ) in multiparous women, and studies on early postpartum (EPP) women with back pain found that 60% had BME visible on MRI.
Misdiagnosis of spondyloarthritis is a matter of concern in pregnant and postpartum women with pain, in whom BME could be merely transient.
What does this study add?
The prevalence of BME was significantly greater among EPP women (33%) than nulliparous (NP) women (14%, p=0.001) but was not significantly different to that of late postpartum (LPP) women (21%, p=0.071).
The Assessment of SpondyloArthritis international Society (ASAS) MRI criteria for sacroiliitis were positive in 25% of the EPP women, 10% of the NP women and 17% of the LPP women, while among the women who had BME, the ASAS MRI criteria for sacroiliitis were positive in 75% of EPP women 71% of NP women and 80% of LPP women.
How might this impact on clinical practice or future developments?
The ASAS MRI criteria for sacroiliitis must be applied within a defined context and interpreted in view of the combinations (BME with erosions and ankylosis) and locations of lesions.
In the context of chronic back pain, MRI of the sacroiliac joints (SIJ) has allowed the description and definition of lesions highly suggestive of sacroiliitis, a major criterion for the diagnosis of non-radiographic axial spondyloarthritis (ax-SpA).1 In fact, the Assessment of SpondyloArthritis international Society (ASAS) guidelines, applicable to patients with back pain aged <45 years at onset of symptoms, classify ax-SpA as ‘an active inflammation of the SIJ visible on MRI, together with one or more other features evocative of SpA’.2 3 MRI enables early diagnosis of ax-SpA, as it reveals active inflammation seen as bone marrow oedema (BME), before progression to structural lesions. The specificity of MRI remains disputed, however, since conditions associated with SIJ lesions can mimic sacroiliitis and lead to overdiagnosis.4 5 Positive diagnosis of SpA can result in inappropriate prescription of biological drugs (anti-tumour necrosis factor α or anti-interleukin-17), which are associated with adverse events and considerable costs.6 Older age, regional anatomic variations and physical strains from intensive sports can all cause BME at the SIJ, evocative of sacroiliitis.
During pregnancy and delivery, the pelvic joints undergo changes caused by hormonal alterations and mechanical stresses, which can result in pelvic or low back pain.7 Recent studies suggested that BME and degeneration are frequently observed at the SIJ in multiparous women,8–10 and studies on early postpartum (EPP) women with back pain found that 60% had BME visible on MRI.9 11 Misdiagnosis of SpA is thus a matter of concern in pregnant and postpartum women with pain, in whom BME could be merely transient.12 13 In a recent prospective cross-sectional study of 204 participants, Seven et al 13 found BME and fat lesions to be most pronounced in patients with ax-SpA, followed by postpartum women with back/pelvis pain. Moreover, images acquired later after delivery revealed fewer lesions, and very few of the mothers who had given birth >16 months prior to MRI acquisition had BME or erosions.13
To the authors’ knowledge, there is only one published study that investigated inflammatory signs at the SIJ, in a large cohort of women at different postpartum periods.13 The purpose of this prospective study was to determine the prevalence of BME at the SIJ in EPP, nulliparous (NP) and late postpartum (LPP) women, and to identify factors associated with the presence of BME at the SIJ. The hypotheses were: (1) that the prevalence of BME is greater in EPP women than in NP or LPP women and (2) that BME diminish over the first 12 months following delivery.
Materials and methods
Design and study population
From November 2015 to May 2019, the authors at four centres prospectively enrolled all women sent by the gynaecology department to acquire a pelvic MRI for reasons that do not trigger SIJ BME and are generally unrelated to musculoskeletal disorders (60% endometriosis, 10% gynaecological cancer, 10% ovarian cysts, 7% uterine myomas…). The study was designed to obtain three groups:
NP women, who had never given birth.
EPP women, who had given birth within the 12 months preceding MRI acquisition.
LPP women, who had given birth more than 24 months prior to MRI acquisition.
The inclusion criteria were: (1) women aged between 18 and 50, (2) availability of MRI with additional Short Tau Inversion Recovery (STIR) or fat-sat T2-weighted sequence to check for a pelvic or gynaecological pathology and (3) national health insurance coverage. The exclusion criteria were: (1) family history of inflammatory disease, (2) patient history of inflammatory disease, (3) childbirth between 12 and 24 months prior to MRI acquisition, (4) pregnancy at the time of MRI acquisition, (5) patient history of pelvic pathology, trauma or surgery. Breastfeeding women were not excluded as MRI did not require any contrast agents.
Each patient completed a questionnaire that collected demographic data (age, weight, height, time from last delivery, total number of deliveries), medical history (antecedents of back/pelvis pain, yes/no) and sports practice. By discussion and consensus among the authors, sports were categorised as either affecting or not affecting the SIJ (table 1).
MRI was performed on 1.5T MRI units (Siemens, Magnetom Aera; Siemens, Area; General Electrics, Optima MR60; Toshiba, Titan) and a 3T MRI unit (Philips, Ingenia). MRI was acquired following to the standard protocol of a pelvis exam, from the iliac crest to the lesser trochanter, in the axial oblique plane with a field of view of 400×400 mm for all sequences. A T1-weighted (slice thickness between 4.2 mm and 6 mm) (TR 619 ms/TE 9 ms) sequence was systematically realised to confirm the absence or presence of structural lesions (erosions, osteophytes and ankylosis), and an axial STIR sequence (2.30 min) (TR 2800 ms/TE 68 ms) or axial T2 fat-sat (TR 3939 ms/TE 116 ms) or axial T2 Dixon (TR 5600 ms/TE 54 ms) was added to the protocol to confirm the absence or presence of any BME or sclerosis. Since the three sequences demonstrated equivalent accuracy,14 each participating centre was allowed to use the sequence of their standard practice.
The SIJ MRI examinations were independently analysed by two musculoskeletal radiologists, who were blinded to clinical and other imaging findings. Readers gave a general dichotomous verdict on a ‘positive ASAS MRI criteria for sacroiliitis’3 (figures 1 and 2). In case of disagreement between the two radiologists, a third senior radiologist was asked to adjudicate cases, and the final verdicts or values were established by consensus.
The primary outcome was the presence of BME and/or structural lesions (erosions, osteophytes, ankylosis and sclerosis) at the SIJ as defined in the ASAS handbook.2 BME was defined as a high signal area within the subchondral bone on MRI STIR/T2 fat-sat/T2 Dixon sequence. Erosions were defined as bony defects at the joint margin, osteophytes as formation of new juxta-articular bone, and ankylosis as fusion of the SIJ, on either MRI T1 or STIR/T2 fat-sat/T2 Dixon sequence. Sclerosis was defined by low-intensity or signal-free bands on both sequences (T1 and STIR or T2 fat-sat or T2 Dixon). Furthermore, a morphological analysis of BME was performed, reporting their extent (mediolateral depth and anteroposterior length) and location (three superior slices, three central slices and three inferior slices). ASAS MRI criteria for sacroiliitis were positive if BME was seen on two successive slices or at two locations on the same slice.
Sample size calculation
Based on the senior radiologist’s experience and preliminary results, the proportion of BME at the SIJ was expected to be greater by 20% for EPP women compared with NP women or LPP women. Considering that during the inclusion period, there were twice as many NP women and LPP women than EPP women, the sample size needed to significantly detect a difference of 20% with a statistical power of 80%, was 297 patients: 59 in the EPP group, 119 in the NP group and 119 in the LPP group.
Intergroup differences were evaluated using Wilcoxon rank-sum and Fisher’s exact tests. Univariable analysis was performed to test associations between the presence of BME and nine independent variables. Multivariable analysis was performed with only four independent variables, excluding those that exhibited strong collinearities with parity (number of deliveries, antecedents of back/pelvis pain, erosions, sclerosis and osteophytes). Intraclass correlation coefficients (ICCs) for continuous data (BME depth and length, and number of slices on which BME was visible) and Gwet’s AC15 for categorical data (presence of BME, erosions, sclerosis, ankylosis and osteophytes, BME location, ASAS MRI criteria for sacroiliitis) were used to assess the level of agreement between the MRI readings of two independent readers (AH and ChL). ICC and Gwet’s AC values can be interpreted as follows:<0.40 poor; 0.40–0.59 fair; 0.60–0.74 good and 0.75–1.00 excellent.16 Statistical analyses were conducted using R V.3.5.2 (R Foundation for Statistical Computing). P values<0.05 were considered statistically significant.
Interobserver agreement (Gwet’s AC) was excellent for the presence of BME (0.76; 95% CI 0.58 to 0.95), BME location (0.90, 95% CI 0.76 to 1.00), ASAS MRI criteria for sacroiliitis (0.92, 95% C.I. 0.76 to 1.00) and presence of erosions (0.90, 95% C.I. 0.79 to 1.00). Interobserver agreement (ICC) was excellent for BME length (0.89, 95% C.I. 0.77 to 0.95) and depth (0.95, 95% C.I. 0.86 to 0.98).
Patient and public involvement section
There was no patient or public involvement in this study.
Of the 517 patients included, 48 were excluded due to family history of inflammatory disease (3 ulcerative colitis, 9 Crohn’s disease, 8 SpA, 12 psoriasis, 14 rheumatoid arthritis and 2 others), 22 due to patient history of inflammatory disease (4 Crohn’s disease, 4 SpA, 7 psoriasis, 1 rheumatoid arthritis and 6 others), 12 due to having given birth between 12 and 24 months prior to MRI acquisition, 6 due to ongoing pregnancy at the time of MRI acquisition and 6 due to history of pelvic pathology, trauma or surgery, leaving 423 patients for analysis (figure 3).
The NP group comprised 219 women aged 30.9±7.7 years with a body mass index (BMI) of 23.1±4.8 kg/m2 (table 1). Fifty-five (25%) women had antecedents of back/pelvis pain and 51 (23%) played sports considered to affect the SIJ. The EPP group comprised 60 women aged 31.2±4.5 years with a BMI of 24.0±4.5 kg/m2. Twenty-five (42%) women had antecedents of back/pelvis pain and 22 (37%) played sports considered to affect the SIJ. Most (75%) women had only one child, while 11 (18%) had 2 children, and 4 (7%) had 3 or more children. The LPP group comprised 144 women aged 38.3±6.2 years with a BMI of 24.0±5.1 kg/m2. Sixty-four (44%) women had antecedents of back/pelvis pain and 27 (19%) played sports considered to affect the SIJ. Fifty-eight (40%) women had only one child, while 41 (28%) had 2 children, and 45 (31%) had 3 or more children.
EPP women were of comparable age to NP women (p=0.360) but significantly younger than LPP women (p<0.001) (table 1). EPP women had fewer deliveries than LPP women (p<0.001). Antecedents of back/pelvis pain were significantly more prevalent among EPP women than NP women (p=0.025) but equivalent to LPP women (p=0.645). Both EPP and NP women played more sports that are considered to affect the SIJ than LPP women (p=0.011, p=0.003, respectively).
MRI findings at the SIJ
The prevalence of BME was significantly greater among EPP women (33%) than NP women (14%, p=0.001) but was not significantly different to that of LPP women (21%, p=0.071) (table 2). The ASAS MRI criteria for sacroiliitis were positive in 15 (25%) of the EPP women, 22 (10%) of the NP women and 24 (17%) of the LPP women. Among the women who had BME, the ASAS MRI criteria for sacroiliitis were positive in 75% of EPP women 71% of NP women and 80% of LPP women.
EPP women had a similar prevalence of erosions (7%) than LPP women (6%, p=1.000) and NP women (5%, p=0.510). It is worth noting that the prevalence of erosions was less than 10% in all three groups, and that BME and erosions were simultaneously present in 7% of EPP women, 5% of LPP women and 2% of NP women. EPP women had a similar prevalence of sclerosis (38%) than LPP women (28%, p=0.135) but a significantly greater prevalence of sclerosis than NP women (18%, p=0.001). Lastly, EPP women had a similar prevalence of osteophytes (28%) than LPP (42%) and NP (27%) women, although there was a significant difference between LPP and NP women (p=0.006).
Morphological analysis of BME at the SIJ
BME length was significantly smaller in NP women (12.0±5.4 mm) than EPP (20.0±11.4 mm, p=0.002) and LPP (17.1±7.6 mm, p=0.006) women (table 3). For women in all three groups, the most common location of BME was central (65%–87%). Most (60%) EPP women had BME at both the sacrum and ilium, while most (52%) NP women had BME only at the sacrum (p=0.040).
Factors associated with BME
Univariable analysis revealed that presence of BME significantly increased with age (OR 1.03 per year; p=0.038) and was higher for EPP women (OR 3.03; p=0.001) compared with NP women. It also revealed that the presence of BME was associated with antecedents of back/pelvis pain (OR 1.86; p=0.014), erosions (OR 11.71, p<0.001), sclerosis (OR 11.12, p<0.001) and osteophytes (OR 5.17, p<0.001) (table 4). Multivariable analysis confirmed that presence of BME significantly increased with age (OR 1.04 per year; p=0.035) and was higher for EPP women (OR 3.68; p<0.001) compared with NP women.
Subgroup analysis of EPP women
Compared with EPP women who had no BME, those with BME had more sclerosis (80% vs 18%; p<0.001) and osteophytes (60% vs 13%; p<0.001), and tended to have more erosions (20% vs 0%; p=0.09). It is interesting to note that the postpartum period was shorter for the 20 EPP women who had BME (3.2±2.5 months) than for the 40 EPP women who did not (4.9±3.3 months, p=0.072), though the difference was not significant, probably due to the small number of observations. A bivariable analysis revealed that the odds of having BME decrease with time from delivery (OR 0.80 per month; 95% CI 0.63 to 0.98; p=0.045), independently from patient age (OR 1.10 per year; 95% CI 0.97 to 1.26; p=0.155).
We investigated MRI findings at the SIJ in NP and postpartum women, and found that (1) BME was present in all three groups, but at a significantly higher prevalence in EPP women, compared with NP women; (2) erosions were present in all three groups, with no significant difference in prevalence between groups; (3) sclerosis was present in all three groups, but at a significantly lower prevalence in NP women, compared with EPP and LPP women; (4) osteophytes were present in all three groups, but at a significantly lower prevalence in NP women, compared with LPP women; (5) ankylosis was not present in any of the groups and (6) in the EPP group, BME decreased monthly in the first year post partum; thus our hypotheses were confirmed.
Women in the EPP group had the highest prevalence of BME (33%), erosions (7%) and sclerosis (38%) compared with NP and LPP women; this is in agreement with that reported by Seven et al 13 who measured a prevalence of BME and erosions of 35% and 7% in 60 postpartum women who had given birth within the preceding 4–16 months. Arnbak et al 17 studied BME and erosions in 1037 patients with low back pain and found a prevalence of 21% and 8%, while Baraliakos et al 18 studied BME in 793 volunteers and found a prevalence of 17%, and Ziegeler et al 19 studied erosions in 485 asymptomatic subjects and found a prevalence of 3%. In addition, the current study has found that women in the LPP group had the highest prevalence of osteophytes (42%), which could be explained by the older age of LPP women, since osteophytes are a degenerative sign of the SIJ.20
There was a trend for LPP women to have a lower prevalence of BME than EPP women. Furthermore, bivariable analysis on EPP women suggested that BME at the SIJ are merely transient and tend to disappear over time. A recent study on 25 women compared MRIs at the SIJ just after vaginal delivery (<10 days) and at 6 months post partum, and found that the number of women with a positive ASAS MRI criteria for sacroiliitis decreased from 64% after delivery to 17% at 6 months post partum.21 Similarly, Seven et al 13 found that women who had given birth more than 16 months prior to MRI acquisition had fewer BME and erosions (5% and 0%, respectively) than those that had given birth within the preceding 4–16 months (35% and 7%, respectively).
Sacroiliitis according to the ASAS definition of a positive MRI was seen in 75% of women with BME at the SIJ, corresponding to 14% of the overall population. Another study reporting MRI results of 7 postpartum women with back pain, found that more than half (57%) had a positive ASAS MRI criteria for sacroiliitis.11 Other studies found that up to 20% of young patients with chronic back pain had a positive ASAS MRI criteria for sacroiliitis.17 22 Therefore, the ASAS MRI criteria for sacroiliitis can lead to false-positive results, and clinicians should be careful when making an SpA diagnosis in postpartum women.
Our study has found that erosions and ankylosis are relatively uncommon in patients without ax-SpA, even though these patients may present other lesions such as BME and sclerosis, this corroborates what has been previously reported by Ziegler et al 19 and Seven et al.13 In contrast, erosions are common in patients with ax-SpA, and they are relatively specific to this disease.13 23 We believe that although low levels of BME and erosions are common in conditions other than ax-SpA, intermediate-to-high levels of BME along with erosions and ankylosis are highly specific for the diagnosis of ax-SpA. The ASAS MRI criteria for sacroiliitis must therefore be applied within a defined context and interpreted in view of the combinations and locations of lesions.
Strengths of the present study are the prospective design, the high number of participants, including women with different postpartum periods, the strict inclusion and exclusion criteria, and the blinded radiographic assessment conducted by two independent readers. Limitations include that women between 12 and 24 months post partum were excluded from this study because it was uncertain whether they should be considered early or late post partum and that the sample size was not sufficient to prove a statistically significant difference in the prevalence of BME at the SIJ between EPP and LPP women.
EPP women have a higher prevalence of BME at the SIJ than NP women and tend to have a higher prevalence of BME compared with LPP women. The presence of BME at the SIJ decreases with time from delivery. Most women with BME at the SIJ had a positive ASAS MRI criteria for sacroiliitis, indicating that the presence of BME as the main criterion can lead to false-positive results. The interpretation of SIJ MRIs should not be made in isolation since age, time from delivery and other factors may outweigh the pertinence of MRI findings.
The authors are grateful to Mo Saffarini for his assistance with manuscript preparation.
Handling editor Josef S Smolen
Contributors We hereby confirm that all authors have participated in the study as follows: AH: study design, data collection, manuscript editing. CL: study design, data collection, manuscript editing. CL: study design, data collection, manuscript editing. PT: study design, data collection, manuscript editing. J-BP: study design, data collection, manuscript editing. NS: study design, data collection, manuscript editing. SR-P: literature review, data interpretation, figures and tables, manuscript writing. CC: study design, data collection, manuscript editing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The study protocol was approved by the medical ethics committee in advance (Montpellier University Hospital: 2015-A01513-46).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.