Zygapophyseal joints were obtained from eight patients with AS (mean age 45 (range 30 to 59); five men, three women). This study includes patients from a larger group for which histopathological observations from the zygapophyseal joints were reported recently 10. In brief, all eight patients had severe kyphosis and were completely ankylosed in the lumbar spine. The mean disease duration of all eight patients was 22.5 years (range 7 to 33 years). Seven of eight patients reported symptoms of nocturnal back pain before surgery. Histologically detectable edema and cellular infiltrates were also searched for in control samples taken from autopsies of 10 patients without AS who died from cardiovascular diseases and had no history of rheumatic diseases.

Surgery was performed as reported in more detail previously 10: we obtained at least one zygapophyseal joint (Figure 1a) from the lumbar spine of each patient with AS. For all eight patients with AS, preoperative MR images of the lumbar spine and the thoracic spine were available for comparative analysis.

Permission for this study was given by the local ethic committee of the Charité Berlin, Campus Benjamin Franklin. All patients gave permission for histopathological analysis of the obtained material.

The histopathological assessment was performed as described recently 11. In brief, the zygapophyseal joints were cut into thin slices (3 to 4 mm) with a saw and fixed in 4% buffered formalin. After decalcification with EDTA, sections 4 to 6 μm thick were prepared and stained with hematoxylin and eosin for the evaluation of bone marrow edema. Bone marrow edema was characterized by the accumulation of eosinophilic fluid in the bone marrow interstitium in between univacuolar fat cells, resembling the morphological findings described in the bone marrow of patients with severe malnutrition 12, during the first couple of weeks following stem cell transplantation 13 or after cytoreductive therapy in chronic idiopathic myelofibrosis 14 (Figure 1b; red arrows, edema; black arrow, vacuoles of fat cells). The relative amount of edema, calculated as the percentage of the total bone marrow area, was quantified as described in an earlier study in which bone marrow edema was assessed for a comparative analysis of magnetic resonance (MR) images and histopathology in osteoarthritis 15. The sections were also examined by polarized light after staining with Congo red to rule out the presence of amyloid mimicking extracellular edema.

The evaluation of all sections was performed by an experienced pathologist (CL) who was blinded for patients, controls, and the MRI results. Sections were examined with a microscope allowing light and immune fluorescence microscopy (BX60; Olympus, Hamburg, Germany). Pictures were taken with a digital camera (Color View II; Soft Imaging System GmbH, Münster, Germany) and were further analyzed by Analysis^® software (Soft Imaging System).

Immunohistochemistry was performed, as described in more detail recently 11, to detect CD3⁺ T cells, CD4⁺ T cells, CD8⁺ T cells and CD20⁺ B cells. In brief, sections were immersed in a sodium citrate buffer solution at pH 6.0 and heated in a high-pressure cooker (CD3, CD4, CD8 and CD20). After cooking, the slides were incubated with the respective primary antibodies including monoclonal antibodies against CD4 (clone 1F6; Novocastra, Newcastle, UK), CD34 (QBend10; Immunotech, Marseille, France), and CD3 (F7.2.38), CD8 (C8/144B) and CD20 (L26) obtained from Dako (Glostrup, Denmark). For detection, the alkaline phosphatase/anti-alkaline phosphatase complex (APAAP) method was used, with 'Fast Red' as chromogen.

T-cell and B-cell aggregates were defined as clusters of 50 or more CD3⁺ T cells or CD20⁺ B cells, and for each patient the number of CD3⁺ T-cell aggregates or CD20⁺ B-cell aggregates in 10 high-power microscopic fields (HPFs), based on a HPF of 0.237 mm² first, ocular with a 22 mm field of view at ×10 magnification, and a 40× objective), was assessed. For the quantification of interstitial CD4⁺ and CD8⁺ T cells and CD20⁺ B cells in the bone marrow, 10 HPFs were analyzed similarly. The number of interstitial CD4⁺ and CD8⁺ T cells was defined as increased if at least ten CD4⁺ T cells and at least six CD8⁺ T cells per HPF were present; for CD20⁺ B cells the criterion was at least five positive cells per HPF on the basis of the findings in zygapophyseal joints of patients without AS, which we used as negative controls.

The results are given semi-quantitatively: +++, at least two CD3⁺ lymphocytic aggregates and increased interstitial CD4⁺, CD8⁺ and CD20⁺ lymphocytic infiltrates; ++, one CD3⁺ lymphocytic aggregate and increased interstitial CD4⁺, CD8⁺ and CD20⁺ lymphocytic infiltrates; +, increased interstitial CD4⁺, CD8⁺ and CD20⁺ lymphocytic infiltrates, no aggregates. For the quantification of CD34⁺ micro-vessel density a semi-quantitative analysis was performed: 0, not more than two micro-vessels per HPF; +, three or four micro-vessels per HPF; ++, five or six micro-vessels per HPF.

MRI, including T1-weighted, T2-weighted and TIRM (turbo inversion recovery magnitude) sequences, was performed as a preoperative procedure in patients with AS, as shown in Table 1. TIRM sequences allow excellent illustration of the anatomy and detection of inflammation in the bone marrow. The entire lumbar spine including the zygapophyseal joints and, if available, the thoracic spine was analyzed for increased signals in T2-weighted sequences.

We summarize the results of a semi-quantitative analysis of mononuclear cell infiltration in the bone marrow of zygapophyseal joints as described in the Method section. Five patients with AS had the highest score of cellular infiltration (+++) with at least two CD3⁺ lymphocytic aggregates including increased interstitial CD4⁺, CD8⁺ and CD20⁺ lymphocytic infiltrates, and three patients with AS had an increased level of cellular infiltrations (+) without aggregates (Table 1).

To address the question of whether hypervascularization might contribute to the detection of bone marrow edema in the spine by MRI we also stained CD34⁺ endothelial cells in our sections. This issue was examined more thoroughly in a previous paper 10. The present results are shown in Table 1.

All patients with AS showed different degrees of interstitial bone marrow edema in histopathological analysis: in two patients with AS, interstitial edema was present in 10% of the total bone marrow area, in two patients it was present in 20%, in two patients in 30% and in two patients in 60% (Table 1). In contrast, none of the controls showed interstitial bone marrow edema.

Interstitial bone marrow edema as detected by histopathology in the zygapophyseal joints of the lumbar spine was found by MRI in three of eight patients with AS (Table 1). Most interestingly, in these patients a good correlation between histopathologically confirmed interstitial edema and cellular infiltration was found. Two patients showed the highest score for interstitial bone marrow edema (60%) and for mononuclear cell infiltrates (+++). An example of such an AS patient is shown in Figure 2: bone marrow edema as detected by MRI (Figure 2a) was correlated with interstitial bone marrow edema as detected by hematoxylin and eosin staining (Figure 2b) and T-cell infiltration (Figure 2c). The third patient with AS (no. 3) with bone marrow edema detected by MRI had 30% interstitial bone marrow edema and the highest score of bone marrow infiltation in the histopathological assessment. Some of the patients had inflammatory lesions in other parts of the spine, as indicated in Table 1.

Interestingly, histopathologically detected cellular infiltration and interstitial bone marrow edema was also observed, although to a smaller degree (between 10% and 30%), in patients with negative MRI. An example is shown in Figure 3: interstitial bone marrow edema (20%) (Figure 3b) and cellular infiltration (Figure 3c) was detectable in histopathological observations but not by MRI (Figure 3a). In two of the patients in whom zygapophyseal joints were MRI-negative, signs of osteitis were seen at other sites of the spine: in the vertebral bodies of thoracic vertebrae 11 and 12 of patient 4 and at posterior sites of the vertebral bodies of thoracic vertebrae 6 to 9 including the zygapophyseal joints of patient 6 (Table 1).