The findings that separate these two diagnoses are visible on plain radiography. They were not visible on either lumbar study because the relevant anatomy lies elsewhere in the axial skeleton.

The differential here turns on study design, not interpretive skill.

Why the clinical picture doesn’t resolve it

Axial stiffness, pain, and reduced range of motion occur in both conditions, and neither presentation reliably points one way at first contact. In patients over 45, where spondylosis is common and often read as incidental, the inflammatory question may never be raised — particularly when regional imaging shows changes that fit the patient’s age.

Laboratory testing narrows the differential without closing it. HLA-B27 has about 90% sensitivity for ankylosing spondylitis and 6–8% prevalence in the general population;¹ its absence does not exclude the diagnosis. CRP and ESR are often normal in axial SpA, particularly in stable or early disease. History can suggest inflammatory versus mechanical pain but is too imprecise to settle the question on its own — the structural distinction has to come from imaging.

Plain radiography is the first-line structural investigation because it captures distributional and morphologic information that no other modality matches for cost, accessibility, and yield at this point in workup. The information actually delivered depends on which regions were imaged.

What each condition actually looks like

The classical presentations are radiographically distinct. The diagnostic problem is the middle range: moderate disease in both conditions, in patients whose age and metabolic profile make either plausible.

DISH is defined by the Resnick criteria: flowing ossification along the anterolateral aspect of at least four contiguous vertebral bodies, preserved disc height at the involved levels, and no sacroiliac erosion or apophyseal joint ankylosis.² The ossification is thick, exuberant, and characteristically right-sided in the thoracic spine, with the left-sided aorta inhibiting ossification on the opposite side. Disc spaces are preserved because the process is ligamentous and entheseal, with the disc itself uninvolved.

Ankylosing spondylitis produces thin, vertical syndesmophytes arising from the vertebral endplate margins — in advanced disease, the bamboo spine. Earlier, the changes are subtler: anterior corner erosions (Romanus lesions), reactive sclerosis at corners (shiny corner sign), vertebral body squaring from erosion of the normally concave anterior surface. Sacroiliac involvement is present in virtually all patients with established axial SpA — erosion, sclerosis, and eventual ankylosis, bilateral but not always symmetric.

Coexistence of both conditions in the same patient, while uncommon, has been reported.

The grey zone is real: bridging at three levels rather than four, ossification moderately thick but not classically flowing, disc height partially reduced. This describes a predictable share of the clinical population — the same group in which plain radiography is asked to do the most diagnostic work and in which regional imaging is least equipped to deliver it.

Why a single region fails

The thoracic spine isolated

Flowing ossification on an AP thoracic view, without lateral projection and without cervical or lumbar context, cannot be reliably assigned to DISH or AS. Ossification thickness, the relationship to the disc space, the laterality — these may be partially visible on a lateral thoracic view. The level count, disc height across all involved segments, and SI joint status are outside the frame. Right-sided thoracic ossification raises the possibility of DISH; it does not confirm it.

The lumbar spine isolated

Moderate bridging at L2–L4 with preserved disc height is compatible with DISH, with hypertrophic spondylosis, and with early ankylosing spondylitis where thoracolumbar involvement precedes classical lumbar syndesmophyte formation. A lumbar-only study can describe these findings accurately without distinguishing among them, generating a report that is technically correct and clinically inert.

The cervical spine isolated

Anterior bridging at multiple cervical levels may represent DISH, hypertrophic spondylosis, or late AS with cervical extension. DISH and AS produce their most overlapping morphology at cervical levels — the right-sided lateralization characteristic of thoracic DISH is less pronounced, and syndesmophyte thickness in late-stage AS can approach that of DISH ossification. The cervical spine alone is the least informative starting point for this differential.

In each case, the report is accurate for the study performed and inadequate for the clinical question.

What the full axial set reveals

1. Level count and continuity. The four-level DISH threshold requires seeing enough spine to count. Three levels of bridging on a lumbar study may be three of six, or three of three — that distinction is not recoverable without thoracic coverage. Contiguous versus skip distribution, thoracolumbar junction concentration versus diffuse spread — none of this is visible in a two-region study.
2. Ossification morphology across segments. Right-sided thoracic predominance is a distributional feature of DISH. It requires thoracic imaging to identify. Thin, symmetric, marginal syndesmophytes are best characterized on lateral views at multiple levels. Neither finding is reliably characterizable on a study that covers one or two spinal regions.
3. Disc height across all bridged levels. Preserved disc height is the single most reliable DISH criterion — but it must hold across all bridged segments. Selective preservation, with reduction at some bridged levels and not others, changes the interpretation substantially. That assessment requires all bridged segments to be in the same study.
4. Sacroiliac joint status. This is the decisive variable. DISH spares the SI joints by definition; AS involves them in virtually every established case. A single AP pelvis with dedicated SI evaluation converts an ambiguous spinal study into a directed differential. Normal SI joints alongside extensive spinal bridging and preserved disc height makes DISH the overwhelming diagnosis. Erosion, sclerosis, or early ankylosis at the SI joints — regardless of how ‘degenerative’ the lumbar spine appeared in isolation — places the patient in the SpA differential and defines the next step.

The clinical stakes

A patient with DISH labeled as possible axial SpA enters a pathway that serves them badly: rheumatology follow-up, serial MRI, HLA-B27 testing, eventual discussion of biologics. DISH does not respond to NSAIDs the way axial SpA does, and there is no evidence base for biologic therapy in DISH. The cascade following an incorrect SpA diagnosis represents sustained misallocation of clinical attention and avoidable treatment risk.

The opposite error has its own cost. Historical delay from symptom onset to ankylosing spondylitis diagnosis is 6–10 years,³ reflecting nonspecific early symptoms and the routine failure of regional imaging to capture distributional findings. Structural damage accumulates over that interval. The window for biological intervention — most effective before significant ankylosis — narrows and may close. Patients accumulate lumbar studies, physiotherapy referrals, and analgesic prescriptions while the diagnosis that would change management never appears in the record.

Fragmentation obscures the diagnosis. Each partial study, read alone, fits a degenerative pattern; the cumulative record looks adequate while concealing the finding that would change management.

The documentation problem is specific: each report is technically defensible, but no single report contains the variables needed to settle the diagnosis. The record appears adequate to a later reviewer because the inadequacy is distributed across studies, not concentrated in any one of them.

What to do next: the triage logic

DISH pattern is clear: flowing ossification across ≥4 contiguous levels, right-sided thoracic predominance, preserved disc height throughout, normal SI joints, no corner erosions, no squaring. MRI adds nothing to the structural diagnosis here. Management addresses symptom control and metabolic comorbidity. Repeat imaging follows clinical change, not a fixed schedule.

SI joint involvement is present: erosion, asymmetric or progressive sclerosis, joint-space irregularity, or early ankylosis — any of these places the study in the SpA differential. Next step: MRI of the sacroiliac joints, STIR and T1, dedicated SpA protocol, to assess for bone marrow edema as evidence of active inflammation. This is the NICE NG65 pathway: plain films first; MRI when plain films are negative but suspicion persists, or to characterize ambiguous plain film findings.⁴

Pattern is ambiguous or mixed: bridging that doesn’t clearly meet DISH criteria, partially reduced disc height, bilateral SI sclerosis without definite erosion. CT of the sacroiliac joints answers a different question than MRI — it supplies osseous detail, cortical definition, and erosion geometry that plain films cannot resolve in equivocal cases.

Applying this triage logic is straightforward when level count, disc height across bridged levels, ossification morphology, and SI joint status appear in the same study. A fragmented record cannot support the pathway because the relevant variables are scattered across regions that may never share a report.

Study design as a diagnostic act

The radiographic distinction between DISH and ankylosing spondylitis is not primarily an interpretive problem. The features that separate them are visible on plain radiography when the study is designed to capture them. Level count, disc height across all bridged segments, ossification distribution, SI joint status, vertebral corner morphology — these are standard radiographic descriptors that require appropriate regional coverage to assess.

When the study covers only a symptomatic region, the features that would separate the two diagnoses are absent from the imaging set. The report describes the study faithfully while leaving the clinical question unanswered.

Choosing the right study is therefore part of the diagnostic act. A complete axial set with SI joints and lateral projections delivers the variables that separate two conditions whose natural history, treatment, and prognosis diverge sharply. A symptom-driven lumbar series delivers a description of the most painful segment.

These are not equivalent studies. For a patient with bridging axial ossification and an unresolved differential, only the complete axial set carries the information to settle the question.