Clinical and imaging efficacy of infliximab in HLA–B27–Positive patients with magnetic resonance imaging–determined early sacroiliitis

Authors

Errata

This article is corrected by:

  1. Errata: Erratum: Clinical and imaging efficacy of infliximab in HLA–B27–positive patients with magnetic resonance imaging–determined early sacroiliitis Volume 62, Issue 10, 3005, Article first published online: 29 September 2010

  • European Clinical Trials (EudraCT) Database: 2004-001880-23, DDX No MS8000MN/13157.

Abstract

Objective

To evaluate the efficacy of infliximab in HLA–B27–positive patients with magnetic resonance imaging (MRI)–determined early sacroiliitis, using both clinical and MRI assessments.

Methods

Forty patients with recent-onset inflammatory back pain, as assessed by the Calin criteria, HLA–B27 positivity, clinical disease activity as measured by the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), pain and morning stiffness, and magnetic resonance imaging (MRI)–determined sacroiliac joint bone edema were randomized in a double-blind manner to receive infliximab 5 mg/kg or placebo at 0, 2, 6, and 12 weeks. MRI scans were performed at baseline and 16 weeks and scored by 2 observers (blinded to both the order of the scans and to treatment group), using the Leeds scoring system. Clinical assessments included the BASDAI, the Bath Ankylosing Spondylitis Functional Index (BASFI), the Ankylosing Spondylitis Quality of Life (ASQoL) instrument, the ASsessment in Ankylosing Spondylitis International Working Group criteria (ASAS) for improvement, and markers of inflammation.

Results

The mean reduction in the total MRI score from week 0 to week 16 was significantly greater in infliximab-treated patients compared with placebo-treated patients (P = 0.033). On average, significantly more lesions resolved in the infliximab group (P < 0.001), while significantly more new lesions developed in the placebo group (P = 0.004). Significantly greater improvement in the infliximab group versus the placebo group was also observed for changes from week 0 to week 16 in the BASDAI (P = 0.002), BASFI (P = 0.004), and ASQoL (P = 0.007) scores. Responses according to the ASAS criteria for 40% improvement, the ASAS criteria for 20% improvement in 5 of 6 domains, and ASAS partial remission were achieved by 61%, 44%, and 56% of infliximab-treated patients, respectively. Infliximab was well tolerated, and no serious adverse events were observed.

Conclusion

Infliximab was an effective therapy for early sacroiliitis, providing a reduction in disease activity by week 16. This study is the first to show that infliximab is effective for reducing clinical and imaging evidence of disease activity in patients with MRI-determined early axial spondylarthritis.

Ankylosing spondylitis (AS), the prototype spondylarthritis (SpA), is increasingly recognized as an important and potentially treatable condition. AS is more common than previously estimated, with some studies suggesting a prevalence similar to that of rheumatoid arthritis (1). Importantly, AS affects individuals at a time when they are economically active (most commonly in the third decade of life), and the disease has a major impact on their ability to work (2).

Anecdotal evidence has suggested that patients who are treated early, i.e., before radiographic changes are apparent, have a better outcome (3). However, the current diagnosis of AS according to the modified New York criteria (4) relies on radiographically detectable damage. Therefore, the definitive diagnosis of AS is often delayed by up to a decade, because this is the length of time that may be required for structural changes to develop (5).

The limitations of radiography have been partially addressed by the advent of fat-suppressed magnetic resonance imaging (MRI), which can be used to diagnose axial SpA at a stage when radiographic findings remain normal (6). Histologic studies have confirmed that these early MRI changes are caused by osteitis at the sacroiliac joints (7, 8) and typically evolve into radiographic AS over several years (9).

One of the great challenges in caring for patients with AS is early and accurate diagnosis. The Calin criteria for diagnosis of early inflammatory back pain (10) lack sensitivity and specificity (11, 12). A recently proposed modification of the clinical criteria (13) has been shown to have much greater sensitivity and specificity but has not been tested in early disease. In addition, the serum level of markers of inflammation are not consistently elevated in these patients and thus are not reliable surrogate markers for diagnosis (14, 15).

Based on the strong association between HLA–B27 positivity and AS, testing patients with early inflammatory back symptoms for HLA–B27 is useful for early diagnosis as well as for predicting persistent disease that may be associated with ocular and cardiac manifestations (16). Indeed, the combination of the Calin criteria for inflammatory back pain, HLA–B27 positivity, and MRI-defined early sacroiliitis identifies a group of patients who could experience progression to AS in >90% of cases (17).

The advent of anti–tumor necrosis factor (anti-TNF) antibodies in the field of rheumatology has provided new therapeutic options for patients with AS. The efficacy of anti-TNF therapies in AS has been confirmed in numerous trials (18–21). However, until now, there has been no evidence to justify treatment of MRI-determined early axial SpA with anti-TNF agents, because all of the randomized controlled trials conducted to date (18–21) have used diagnosis according to the definite modified New York criteria (4) for inclusion. In the present study, we combined the Calin criteria for inflammatory back pain, HLA–B27 positivity, and MRI-defined early sacroiliitis to identify a homogeneous cohort of patients with early axial SpA in whom disease would likely progress to AS and randomized these patients to receive infliximab therapy or placebo, in order to ascertain the effectiveness of anti-TNF therapy in patients with preradiographic axial SpA.

PATIENTS AND METHODS

The trial was conducted at the Leeds Teaching Hospitals Trust, Leeds, UK. The local ethics committee approved the study, and written informed consent was obtained from all patients prior to screening.

Inclusion and exclusion criteria.

Eligible patients had a combination of inflammatory back pain, according to the Calin criteria (10), for the previous 3 months to 3 years (Table 1), were positive for HLA–B27, and had MRI evidence of sacroiliitis. All eligible patients had active disease, as defined by at least 2 of the following: a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) score of ≥40, a pain score of ≥40 on a 100-mm visual analog scale, and early morning stiffness lasting ≥45 minutes. From studies of infliximab treatment in established AS with MRI outcome (22), we calculated that at the 5% significance level with 80% power detecting a 50% difference between the placebo and treatment groups in the mean changes from baseline to the end point, 20 patients were required in each group.

Figure 1.

Magnetic resonance images of the sacroiliac joints of a patient before (A) and after (B) receiving infliximab therapy. The extensive bone marrow edema that was observed in both sacroiliac joints before treatment was resolved in the posttreatment scan.

Table 1. Working definition of inflammatory back pain, according to the criteria described by Calin et al (ref.10)*
Question/responsePoints
  • *

    A score of ≥4 indicates the presence of inflammatory back pain.

Age at onset of back discomfort, years
 ≥400
 <401
Onset
 Insidious1
 Not insidious0
Persistence of back discomfort, months
 <30
 ≥31
Associated with morning stiffness
 No0
 Yes1
Response to exercise
 Improves1
 No improvement0

To identify 40 patients with MRI-determined sacroiliac joint bone edema, 49 patients with symptoms and HLA–B27 positivity were screened and underwent MRI. The MRIs were reviewed by musculoskeletal radiologists for the presence of sacroiliac joint bone marrow edema representative of active sacroiliitis. Nine patients had no evidence of sacroiliac joint bone edema on MRI and were therefore not randomized into the study. Other exclusion criteria were past or current tuberculosis, congestive heart disease, treatment with systemic glucocorticoids in the previous month, and treatment of the sacroiliac joints with intraarticular glucocorticoids within the previous 3 months.

Treatment protocol.

Patients fulfilling the inclusion criteria were randomized to receive infusions of placebo or infliximab (5 mg/kg) at 0, 2, 6, and 12 weeks. Treatment with oral nonsteroidal antiinflammatory drugs remained unchanged throughout the study. Intraarticular steroid therapy was not allowed, nor was the introduction of oral glucocorticoids or disease-modifying antirheumatic drugs.

Study assessments.

The efficacy of therapy was determined by evaluating changes in MRI and clinical/functional assessments. The primary study end point was the change in the total MRI score (sum of the 8 sacroiliac joint regions) from week 0 to week 16. Safety was assessed by monitoring patients for the occurrence of adverse events and by collecting samples for clinical laboratory testing at each visit.

MRI protocol and scoring.

MRI scans were performed at baseline (week 0) and week 16, using the commercially available 1.5T Gyroscan ACS-NT system (Philips, Best, The Netherlands). Patients were examined in the supine position, with a flat-surface coil positioned under the sacrum. T1-weighted turbo spin-echo (TSE) and STIR TSE fat-suppressed coronal oblique sequences of the sacroiliac joints and T2-weighted spectral presaturation with inversion recovery sagittal sequences of the whole spine were obtained. The MRI parameters were as follows: for T1-weighted oblique sequences of the sacroiliac joint, repetition time (TR) 892 msec, time to echo (TE) 14 msec, matrix 384 × 512 pixels, field of view (FOV) 320 mm, slice thickness 4.0 mm, slice gap 0.4 mm, number of signals averaged (NSA) 3, and acquisition time 5 minutes 44 seconds; for STIR TSE, TR 2,500 msec, TE 10 msec, matrix 382 × 512 pixels, FOV 320 mm, slice thickness 4.0 mm, slice gap 0.8 mm, NSA 2, and acquisition time 4 minutes 35 seconds.

The pretreatment and posttreatment scans were scored together simultaneously by 2 independent observers (DM and LCC) who were blinded to both the order of the scans and to treatment group, using a previously reported scoring system (3,22). Observer disagreement was resolved by consensus. Regions of MRI-determined sacroiliitis, as defined on T2-weighted fat-suppressed images as bone marrow edema (identified by a high or intermediate marrow signal), were scored. For scoring purposes, the sacroiliiac joint was split into 8 regions, as previously described (22). Briefly, a line was drawn across the upper third of the sacroiliac joint, and the upper right iliac, upper right sacral, upper left sacral, and upper left iliac regions were evaluated. The 4 corresponding lower regions were similarly evaluated. Each region was scored for bone edema as follows: 0 = absent, 1 = mild, 2 = moderate, and 3 = extensive. The scores were based on both the extent of osteitis and its severity (or intensity). An overall score of inflammatory activity was calculated as the sum of scores for bone marrow edema, for a total possible maximal score of 12 per joint (24 per patient). To assess the degree of change between scans, paired scoring was performed for every lesion using this system with, for example, resolution of a severe lesion improving from a score of 3 to a score of 0. In the spine, bone marrow lesions were defined as described above and were recorded as present or absent in the vertebral body, posterior elements including spinous processes and facet joints, and paraspinal soft tissues. A total count of lesions per spinal area for each patient was performed. Change was assessed as resolution of the baseline lesions.

Radiographic assessments.

Radiography of the sacroiliac joints was not routinely performed as part of the study protocol. However, the majority of patients did undergo radiography of the sacroiliac joints just prior to being screened for the study. These radiographs were reviewed by independent radiologists/rheumatologists who were blinded to the treatment group and MRI findings.

Clinical and functional assessments.

At each visit, disease activity was assessed with the BASDAI (23), the physician's global assessment of disease activity, and the swollen and tender joint counts (n = 28) to assess peripheral involvement. The Bath Ankylosing Spondylitis Functional Index (BASFI) (24), the disability index of the Health Assessment Questionnaire (HAQ) (25), and the Ankylosing Spondylitis Quality of Life instrument (ASQoL) (26) were used to assess function, disability, and quality of life at each visit. Also at each visit, spinal mobility was assessed by the measurement of chest expansion and the Schober test of lumbar flexion. Samples were collected at each visit for determination of the erythrocyte sedimentation rate (ESR) and the C-reactive protein (CRP) level, both of which are markers of inflammation. The ASsessment in Ankylosing Spondylitis International Working Group criteria for 40% improvement (ASAS40) (27), 20% improvement in 5 of 6 domains (27), and for partial remission (28) were applied to these clinical data.

Statistical analysis.

Data were summarized and tested according to distribution and type. Means and standard deviations are presented for normally distributed continuous data, and Student's t-tests were used to compare the 2 treatment groups. Medians and interquartile ranges were presented for non-normally distributed and ordinal data, which were compared with nonparametric Mann-Whitney U tests. Nominal data were summarized as the number (%) and compared with Pearson's chi-square tests, continuity-corrected as appropriate if the expected cell count dropped below 5. All tests were performed at the 5% level of significance. Corrections for multiple testing within each family of statistical tests were applied for secondary outcomes, according to Holm's modification of the Bonferroni adjustment. The following thresholds for statistical significance of secondary outcomes were derived: for the Mann-Whitney U test, 0.007; for Student's t-test, 0.025; for the chi-square test, 0.017. All analyses were performed using SPSS software version 15.1.1 (SPSS, Chicago, IL).

RESULTS

Patient demographics.

Forty patients were randomized (20 to infliximab treatment and 20 to the placebo group). One patient (in the placebo group) withdrew consent at 12 weeks and discontinued participation in the study. This patient was included in the analysis of efficacy, because it was performed on an intent-to-treat basis. The remaining 39 patients completed the study. The randomized treatment groups were similar with regard to baseline demographic and disease characteristics (Table 2). Despite an average duration of back pain of ∼1–1.5 years, patients had evidence of active disease (mean BASDAI score >5.5), causing significant impairment of functional ability and quality of life. Associated extraarticular features included iritis in 6 patients (15%), psoriasis in 4 patients (10%), and inflammatory bowel disease in 1 patient (2.5%). Three patients (7.5%) had a family history of AS, 6 patients (15%) had a family history of psoriasis, and 4 patients (10%) had a family history of inflammatory bowel disease.

Table 2. Baseline characteristics of the patients*
CharacteristicPlacebo (n = 20)Infliximab (n = 20)
  • *

    Except where indicated otherwise, values are the median. For C-reactive protein (CRP), n = 18 patients in the placebo group and 19 patients in the infliximab group. For the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) score, n = 19 patients in the placebo group. No significant differences between placebo and infliximab were observed for any of the variables. BASFI = Bath Ankylosing Spondylitis Functional Index; HAQ = Health Assessment Questionnaire; ASQoL = Ankylosing Spondylitis Quality of Life; NSAID = nonsteroidal antiinflammatory drug; SI = sacroiliac; MRI = magnetic resonance imaging.

No. of men/no. of women15/515/5
Age, mean years28.229.5
Mean duration of back pain, months13.417.2
Mean BASFI score4.114.42
Mean BASDAI score5.765.85
CRP, mg/liter11.55.0
HAQ score (range 0–3)0.880.88
ASQoL score (range 0–18)11.0010.00
Chest expansion, mm5.005.00
Lumbar flexion, mm4.003.750
Concomitant NSAID use, no. (%) of patients18 (90)18 (90)
Mean number of SI joint lesions on MRI3.13.65

MRI findings.

Thirty-nine patients (20 in the infliximab group and 19 in the placebo group) had MRI performed at baseline and week 16. One patient did not have a repeat scan at 16 weeks because he had withdrawn consent at 12 weeks; the last observation carried forward was used for analysis in his case. All patients had lesions on baseline MRI scans consistent with active disease, when reviewed by musculoskeletal radiologists prior to entry into the study. When blinded paired scoring of the scans was performed after the study, 3 patients (all in the placebo group) had a baseline MRI score of 0. The remainder of the patients all had scores of grade 1 or higher. Seventy percent of the patients had bilateral sacroiliitis. The median total MRI score for the sacroiliac joints at baseline was 3.5 (interquartile range [IQR] 2, 8), and the median changes at week 16 were 0 (IQR –2.00, 1.50) in the placebo group and −2.00 (IQR −6.25, 0.00) in the infliximab group (for the primary end point, Mann-Whitney U asymptotic z = −2.17, P = 0.033). None of the patients in the infliximab group had an increase in the total MRI score, whereas 5 patients in the placebo group did have such an increase.

At baseline, 75 and 60 abnormal sacroiliac joint lesions were observed in the infliximab group and the placebo group, respectively. Among the abnormal baseline joint regions (score ≥1), 47 (62.7%) in the infliximab group resolved completely, versus 20 (29.4%) in the placebo group (Pearson's χ2 = 15.84, 1 df, P < 0.001). When considering joint regions with moderate or high levels of inflammation (MRI score ≥2) at baseline (24 in the placebo group and 22 in the infliximab group), 6 (22.2%) in the placebo group resolved completely, compared with 17 (77.3%) in the infliximab group (Pearson's χ2 = 14.75, df = 1, P < 0.001).

Considering those regions with no MRI abnormality at baseline (MRI score = 0), 92 sacroiliac joint regions in the placebo group (57.5%) were normal, compared with 85 sacroiliac joint regions in the infliximab group (53.1%) (P = 0.431, by chi-square test). Among these normal joint regions, new lesions developed in 11 (12.0%) in the placebo group compared with 1 (1.2%) in the infliximab group (Pearson's χ2 = 8.12, 1 df, P = 0.004).

Patients with moderate or extensive bone marrow edema (at least 1 sacroiliac region with a score of grade ≥2) were compared with those with mild edema only (all regions scoring grade 1). In the placebo group, there was no significant difference in the MRI response (change in total lesion score) between patients with mild baseline bone edema (median 1, IQR −2, 3) and those with moderate or extensive edema (median −0.5, IQR −5.25, 0; P = 0.475, by Mann-Whitney U test). In the infliximab group, patients with mild baseline bone edema showed significantly less improvement in the total MRI score (median 0.0, IQR −2.0, 0.0) than those with moderate or extensive edema (median −7, IQR −12, −2; P = 0.002).

Although all patients manifested baseline sacroiliac joint inflammation, only 9 patients (22.5%; 5 in the infliximab group and 4 in the placebo group) had inflammation detectable on spinal MRI scans. Among patients with spinal inflammation, an average of 3 (range 1–6) spinal lesions were present. Three of the 5 infliximab-treated patients with spinal lesions had complete resolution (10 of 12 lesions resolved), compared with 1 of 4 patients in the placebo group (4 of 14 lesions resolved) (P = 0.016).

Plain radiography findings.

In 34 of 40 patients, plain radiographs of the sacroiliac joints were obtained at baseline. Sixteen (47%) of the 34 radiographs were entirely normal, 14 (41%) of the 34 radiographs showed minor abnormalities not sufficient to make a diagnosis according to the modified New York criteria, and 4 (12%) of the 34 radiographs fulfilled the modified New York criteria for a radiographic diagnosis of AS.

Results of clinical and functional assessments.

The week 16 findings for clinical and functional assessments are summarized in Table 3. The mean reduction in the BASDAI score at 16 weeks was significantly greater in the infliximab group (−3.41) than in the placebo group (−0.75) (P = 0.002). Functional ability, as assessed by the BASFI score, also improved to a significantly greater extent in the infliximab group (mean reduction −2.70) compared with the placebo group (mean reduction −0.47) (P = 0.004). There was no significant difference between treatment groups in CRP level reductions; however, baseline CRP levels of 11.5 mg/liter in the placebo group and 5.0 mg/liter in the infliximab group likely precluded detection of a treatment-related difference. There was a trend toward a larger reduction in the HAQ disability index score among infliximab-treated patients (−0.44) compared with placebo-treated patients (−0.13), but this difference did not reach significance (P = 0.065). There was, however, significant improvement in quality of life by 16 weeks, as measured by the ASQoL score, in infliximab-treated patients (−6.18) compared with placebo-treated patients (−1.00) (P = 0.007).

Table 3. Results of clinical and functional assessments at week 16*
 Placebo (n = 20)Infliximab (n = 20)P
  • *

    Except where indicated otherwise, values are the median (interquartile range). For the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), chest expansion, and lumbar flexion, n = 19 patients in the placebo group. For C-reactive protein, n = 18 patients in the placebo group and 19 patients in the infliximab group. BASFI = Bath Ankylosing Spondylitis Functional Index; HAQ = Health Assessment Questionnaire; ASQoL = Ankylosing Spondylitis Quality of Life; ASAS40 = ASsessment in Ankylosing Spondylitis International Working Group criteria for 40% improvement.

BASDAI score, mean ± SD−0.75 ± 2.42−3.41 ± 2.530.002
BASFI score, mean ± SD−0.47 ± 2.25−2.70 ± 2.360.004
C-reactive protein, mg/liter−1.5 (−6.6, 0.5)−4.0 (−10.0, 0.0)0.702
HAQ disability index score−0.13 (−0.38, 0.00)−0.44 (−0.93, −0.13)0.065
Chest expansion0.000 (−1.000, 1.500)1.000 (0.000, 1.875)0.095
Lumbar flexion, cm0.00 (−0.50, 1.00)0.50 (0.00, 1.00)0.270
ASQoL score−1.00 (−4.50, 0.75)−6.18 (−10.00, −2.25)0.007
ASAS40 response, no./no. tested (%)3/17 (17.6)11/18 (61.1)0.009
ASAS5/6 response, no./no. tested (%)2/15 (13.3)8/18 (44.4)0.053
ASAS partial remission, no./no. tested (%)2/16 (12.5)10/18 (55.6)0.009

In terms of the composite ASAS response criteria, significantly larger proportions of infliximab-treated patients than placebo-treated patients met the ASAS40 criteria (61.1% versus 17.6%; P = 0.009) for improvement at week 16 (Table 3.) In addition, more than half of infliximab-treated patients (55.6%) met the ASAS partial remission criteria at week 16, compared with 12.5% of placebo-treated patients (P = 0.009) (Table 3).

There was no significant difference in the placebo or treatment arm in changes in the BASDAI score between patients with mild baseline MRI bone edema (score ≤1) and those with moderate or severe baseline MRI bone edema (P = 0.68 in the placebo group and P = 0.71 in the infliximab group). In the infliximab group, a higher proportion of patients with an elevated baseline CRP level (n = 9) achieved all of the ASAS composite measures, compared with those patients (n = 9) with a normal baseline CRP level (for ASAS20, 8 patients versus 6 patients; for ASAS50, 7 patients versus 4 patients; for ASAS70, 6 patients versus 4 patients). However, this difference was not shown to be statistically significant, due to the small numbers of patients in these subgroups.

Adverse events.

Study medication was well tolerated, and no serious adverse events were observed. An allergy to infliximab, characterized by a rash and bronchospasm, developed in 1 patient in the infliximab group. This patient discontinued infliximab therapy at week 6 but continued in the study and had a followup MRI performed at week 16.

DISCUSSION

TNF blockade has been a crucial advance in the treatment of AS in recent years. However, previous studies have been limited to patients in whom the diagnosis of AS was confirmed by the definite modified New York criteria, which rely on the presence of radiographic structural changes, which can take several years to develop. In this study, we used stringent inclusion criteria to identify a homogeneous cohort of HLA–B27–positive patients with MRI-determined sacroiliitis who have early axial SpA and a high likelihood of eventually progressing to radiographically defined AS. In this cohort, infliximab therapy resulted in significant suppression of MRI-determined inflammation relative to placebo, with a significant reduction in the total MRI score, increased resolution of inflammatory lesions, and decreased progression of new inflammatory lesions at week 16 (Figure 1). These results are consistent with those seen in patients with established AS (18, 22, 29). Additionally, we observed impressive clinical responses to infliximab therapy, with improvement in disease activity, functional status, and quality of life seen at week 16.

When comparing our findings with those derived from a study of infliximab in established AS (20), the proportion of patients reaching the ASAS partial remission criteria was far higher (55.6% versus 22.4%), suggesting a clear benefit for early treatment of apparently more reversible disease. Previous studies in established AS have shown that patients with a shorter disease duration have a better response to anti-TNF agents (30), suggesting that very early treatment before damage is evident on radiographs will lead to improved responses. Therefore, this study confirms the efficacy of infliximab in early axial SpA and suggests that these patients may respond more favorably to infliximab treatment than patients with radiographically evident, established AS (18, 22, 29).

As noted, the lack of a significant treatment effect on reductions in CRP levels is likely attributable to a low median baseline CRP level of 5 mg/liter in the infliximab group. It is known that CRP levels can be variable in AS, with some patients showing very little systemic inflammatory response (14, 15). It appears that in the early stages of disease that characterized our study cohort, a higher proportion of patients will have normal levels of markers of inflammation, and that these markers cannot be relied on to aid diagnosis or guide therapy. The results suggested a better response in those individuals with an elevated baseline CRP level; however, patients with a normal CRP level did respond to infliximab, with two-thirds of patients achieving an ASAS20 response.

These patients represent a select group of those presenting with inflammatory low back pain, having a relatively brief symptom duration and active disease both clinically and on MRI. Their disease was impacting their functional ability and quality of life, and no current proven treatment options were available to them. Given that the average age of the patients was younger than 30 years, the majority were working full-time. However, many were struggling to cope because of their disease. Until now, there has been no evidence to guide the treatment of such patients. Current guidelines on the treatment of AS are all based on a diagnosis according to definite New York criteria, which require radiographic evidence of disease (31). This study provides novel data on the clinical and MRI-determined efficacy of TNF blockade in HLA–B27–positive patients with active sacroiliitis. Of note, patients with HLA–B27–positive inflammatory back pain without MRI-defined osteitis were excluded from this study. The role of anti-TNF agents in such patients early in the course of disease requires separate evaluation.

Several recent studies have examined the impact of anti-TNF agents on radiographic changes in AS (32, 33). This research has not provided convincing evidence of slowing of radiographic progression, despite anti-TNF treatment for up to 4 years. Explanatory theories suggest that inflammation and bony progression are 2 independent pathways, and that suppression of inflammation may not prevent spinal ankylosis (34). The impact of TNF blockade on new bone formation in the spine has been controversial, and it has been suggested that once inflammation has started, the bony progression cannot be halted with suppression of TNF alone. If this is the case, early intervention, when only sacroiliitis is present, may prevent the development of osteitis at other sites and subsequent bone growth and fusion at these sites.

Interestingly, despite the short duration of symptoms among patients in this study, 4 patients already met the modified New York criteria for a diagnosis of AS. This has been observed in other cohorts of patients with early inflammatory low back pain (35) and reflects the fact that some patients do have rapidly progressive disease.

A recent study by Haibel et al (36) examined the efficacy of adalimumab in the treatment of patients with axial SpA without radiographically defined sacroiliitis. Although that study showed efficacy for anti-TNF therapy in this group of patients, there are substantial differences in the population treated in that study compared with the population treated in the current study. The mean duration of symptoms in the patients receiving active treatment in the study by Haibel et al (7 years [range 2–16 years]) was significantly longer than that in our cohort and cannot really be regarded as representing early disease. The inclusion criteria for the study by Haibel et al required the presence of 2 of the following 3 features: inflammatory back pain, HLA–B27 positivity, and bone edema on MRI. Of the patients treated with adalimumab in that study, only 14% had the combination of HLA–B27 positivity and abnormal MRI results. The study population had a relatively high proportion of female patients (59%) and a lower prevalence of HLA–B27 positivity (59%), which is consistent with an undifferentiated SpA cohort rather than an AS cohort. For these reasons, we believe that the study by Haibel and colleagues represents treatment of a heterogeneous group of patients with SpA, many of whom would not be expected to experience progression to AS. We believe that patients in our cohort, all of whom were HLA–B27 positive and had abnormal MRI results, are more representative of those with early AS.

In conclusion, this study is the first to demonstrate that infliximab is effective for reducing clinical and imaging evidence of disease activity in a cohort of patients in whom progression to radiographic AS is highly likely. The findings of this study highlight the need for a diagnosis of AS that is not based on radiographic criteria and suggest that there is a place for anti-TNF therapy in early axial SpA. However, because SpA occurs in young people, and the long-term risks of malignancy following anti-TNF therapy are currently unknown, a careful consideration of potential risks and benefits should be made before beginning anti-TNF therapy for any patient. Further research is needed to determine whether this early treatment will halt radiographic progression at the sacroiliac joints and elsewhere. If early treatment of axial SpA will also prevent loss of work participation and result in sustained improvement in the quality of life, these patients merit early intervention.

ROLE OF THE STUDY SPONSOR

Centocor had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Centocor.

AUTHOR CONTRIBUTIONS

Dr. Emery had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Barkham, O'Connor, Fraser, Cawkwell, Emery.

Acquisition of data. Barkham, Keen, Coates, Fraser, Cawkwell, Bennett, Emery.

Analysis and interpretation of data. Barkham, Keen, Coates, O'Connor, Hensor, McGonagle, Emery.

Manuscript preparation. Barkham, Keen, Coates, McGonagle, Emery.

Statistical analysis. Hensor.

Radiographic scoring. Bennett.

Ancillary