A single-center retrospective cohort study conducted at the First Affiliated Hospital of Xinjiang Medical University has found that greater interspinous ligament hyperintensity length on short tau inversion recovery magnetic resonance imaging and the presence of occult cervical instability are statistically associated with persistent axial neck pain at twelve months. Evaluating a well-defined cohort of adult patients receiving routine non-surgical and surgical spine care, researchers sought to determine whether adding detailed soft-tissue fluid measurements and radiographic dynamic translation assessments could clarify long-term recovery trajectories. The primary composite clinical outcome required both a twelve-month numerical rating scale pain score of at least three points and a relative symptom reduction of less than thirty percent from baseline. Across multivariable logistic regression models, each five-millimeter increase in interspinous ligament lesion length among patients with detectable hyperintensity was associated with an adjusted forty-nine percent increase in the odds of experiencing residual pain and inadequate recovery. Concurrently, occult dynamic instability was associated with more than two-and-a-half times higher odds of the unfavorable twelve-month outcome, whereas the binary presence of hyperintensity alone failed to demonstrate statistically significant prognostic discrimination after comprehensive multivariable adjustment.
Chronic axial neck pain is a major contributor to global disability, physical limitation, and healthcare utilization, with burden projections expected to escalate alongside population aging and demographic expansion. While degenerative cervical spondylosis is conventionally evaluated using plain radiographs and standard magnetic resonance imaging, static structural abnormalities such as disc degeneration, osteophytes, and facet arthropathy frequently correlate poorly with patient-reported pain intensity or functional impairment. Clinical guidelines routinely emphasize conservative management pathways, active exercise, patient education, and functional restoration, yet substantial subgroups of patients experience protracted discomfort that fails to resolve over annual follow-up intervals. Prior research demonstrated that high-intensity zones within intervertebral discs and fluid signals within facet joints capture localized mechanical stress and inflammatory changes, but the prognostic value of posterior ligamentous inflammation and subtle dynamic translation has remained unestablished. By focusing on interspinous ligament signal alterations visible on fat-suppressed fluid-sensitive sequences and motion-dependent vertebral displacement captured on functional dynamic radiographs, the investigators aimed to evaluate whether integrating these motion-segment parameters could improve twelve-month outcome risk stratification within an observed routine care pathway.
The study cohort was drawn from consecutive adult candidates presenting with degenerative cervical spondylosis and chronic axial neck pain between January 1, 2021, and December 31, 2024, with formal reporting structured in accordance with the TRIPOD-plus-AI statement guidelines. Chronic pain was defined as localized midline or paramedian neck pain extending from the suboccipital region to the shoulder girdle for at least three months, with symptom intensity meeting an initial numerical rating scale cutoff of at least three points. Among 512 candidate individuals screened, 70 did not meet clinical criteria, 60 lacked paired imaging, 15 had inadequate image quality, 10 lacked key baseline variables, and 32 lacked evaluable twelve-month outcomes, leaving 325 patients. The final analyzed cohort comprised 325 patients whose paired baseline imaging examinations and complete twelve-month medical records were fully documented. Selection analyses demonstrated that excluded clinically eligible patients exhibited modest differences in baseline pain severity and symptom duration compared to included participants, highlighting the importance of evaluating model applicability within clearly defined institutional settings.
Imaging assessments adhered to partially standardized institutional protocols, incorporating sagittal two-dimensional fast spin-echo short tau inversion recovery sequences obtained on either 1.5-Tesla or 3.0-Tesla scanner systems utilizing dedicated sixteen-channel cervical spine coils. Both field-strength protocols maintained an identical echo time of fifty milliseconds, two signal averages, a three-millimeter slice thickness, a zero-point-three-millimeter interslice gap, and a 220-millimeter by 220-millimeter field of view, while repetition time and inversion time varied across systems. Concurrently, upright neutral, maximum-tolerable active-flexion, and maximum-tolerable active-extension lateral radiographs were recorded during the same diagnostic assessment window. Occult cervical instability was formally operationalized using calibrated digital radiographic measurements as an absolute neutral sagittal translation of less than two millimeters combined with an abnormal dynamic translation of at least one millimeter between flexion and extension postures. Measurement reliability was evaluated in a sixty-five-patient stratified substudy evaluated by two independent radiologists blinded to clinical outcomes, demonstrating strong interobserver agreement for short tau inversion recovery hyperintensity detection with a Cohen kappa of 0.907, an intraclass correlation coefficient of 0.994 for total length across the full sample, and an interobserver kappa of 0.852 for occult instability classification.
Quantitative measurement of posterior ligamentous lesions followed rigorous morphological criteria to ensure reproducibility and anatomical precision across cervical motion segments from C2/3 through C7/T1. A lesion was defined as an area of hyperintensity within the interspinous ligament displaying signal intensity exceeding adjacent normal ligamentous tissue across at least two consecutive sagittal slices. Radiologists isolated regions of interest using corresponding T1-weighted and T2-weighted structural acquisitions, systematically excluding adjacent bone margins, blood vessels, epidural cerebrospinal fluid, subcutaneous fat, the nuchal ligament, and phase motion artifacts. When continuous signal alterations crossed adjacent anatomical levels, readers partitioned the abnormal intensity according to distinct interspinous ligament anatomical boundaries, summing non-overlapping effective lengths along the craniocaudal dimension. For patients with discontinuous multifocal hyperintensities, individual segment lengths were measured independently and summed, omitting intervening normal tissue. Negative examinations were scored as zero millimeters. Across the entire cohort of 325 patients, 177 individuals displayed detectable interspinous hyperintensity, exhibiting a median total length of 7.8 millimeters among positive cases, while seventy-six individuals met the criteria for radiographic occult instability.
The study’s primary composite endpoint was specified in a formal statistical analysis plan dated January 15, 2025, prior to outcome data extraction and model execution. Residual neck pain with inadequate improvement at twelve months required both an absolute numerical rating scale pain score of three or higher and a relative symptom reduction of less than thirty percent relative to the patient’s baseline rating. This composite framework was designed to differentiate patients who achieved meaningful relative relief despite mild persistent discomfort from those who experienced ongoing clinical stagnation. At the twelve-month assessment window, which spanned twelve months plus or minus thirty days from the baseline imaging prediction date, 113 patients, representing 34.8 percent of the cohort, met the primary endpoint criteria, whereas 212 patients did not. Secondary outcomes tracked continuous pain scores, raw Neck Disability Index changes, and categorical reductions, with patients undergoing intervening cervical spine surgery categorized by their actual twelve-month functional scores rather than being imputed or censored.
Baseline demographic and clinical profiles revealed that patients who met the primary composite endpoint presented with longer median symptom durations and significantly worse initial symptom severity than those who achieved adequate recovery. Specifically, individuals with persistent symptoms had a median symptom duration of 19.9 months compared to 16.6 months among individuals in the non-event group, alongside a higher baseline numerical rating scale score of 6.1 points versus 5.4 points, and a higher baseline Neck Disability Index score of 29.6 versus 27.6 points. Conversely, age, sex distribution, body mass index, smoking history, high occupational physical load, hypertension, diabetes mellitus, and documented anxiety or depression did not differ significantly between outcome groups. Baseline conservative treatments were also balanced, with approximately forty-eight percent of patients in both groups utilizing analgesic medications and forty-five percent participating in structured physical rehabilitation prior to cohort enrollment.
Baseline radiographic and resonance comparisons demonstrated striking structural contrasts between the outcome cohorts across soft-tissue and dynamic biomechanical parameters. Patients meeting the primary endpoint exhibited a significantly higher prevalence of interspinous ligament hyperintensity at 67.3 percent compared to 47.6 percent among patients with favorable twelve-month recoveries. Positive patients in the persistent pain group demonstrated greater median total lesion length at 9.1 millimeters compared to 6.5 millimeters, higher maximum single-level involvement, and a greater number of involved interspinous levels. Furthermore, occult cervical instability was identified in 35.4 percent of patients meeting the primary outcome compared to only 17.0 percent of those who did not, accompanied by significantly greater maximum dynamic translation and larger segmental angular ranges of motion. In contrast, conventional degenerative magnetic resonance variables, including Modic endplate changes, disc protrusions, facet joint degeneration, number of degenerative levels, and neutral C2–C7 Cobb angles, did not differ significantly between the two clinical outcome groups.
In the primary multivariable logistic regression association model, which adjusted for demographic characteristics, baseline functional scores, psychological comorbidity, and Modic changes, interspinous hyperintensity length and occult dynamic instability retained strong independent prognostic associations with twelve-month clinical failure. Each five-millimeter increment in hyperintensity length among positive patients yielded an adjusted odds ratio of 1.493, with a 95 percent confidence interval spanning from 1.111 to 2.005. Occult instability yielded an adjusted odds ratio of 2.584, with a 95 percent confidence interval ranging from 1.445 to 4.623. Higher baseline pain intensity also remained an independent predictor, exhibiting an odds ratio of 1.982 per one-point increase. However, the direct comparison contrasting a centered 7.5-millimeter hyperintensity against no detectable hyperintensity yielded an odds ratio of 1.708 with a 95 percent confidence interval of 0.996 to 2.927, failing to achieve statistical significance. Exploratory spline testing did not demonstrate significant non-linearity for continuous predictors, and interaction tests between occult instability and ligament hyperintensity parameters were non-significant after Holm multiple-testing corrections.
Prediction model discrimination, calibration, and decision-analytic utility were rigorously evaluated through internal validation using twenty repetitions of stratified ten-fold cross-validation coupled with 500 patient-level bootstrap resamples with complete refitting. The combined prediction model achieved an internally validated area under the receiver operating characteristic curve of 0.711, compared to 0.647 for the basic clinical model, 0.691 for the short tau inversion recovery model, 0.683 for the occult instability model, and 0.642 for an extended reference model incorporating routine degenerative findings. However, after applying Holm corrections for multiple testing across model contrasts, none of the incremental gains in discrimination achieved statistical significance, with the combined model versus basic model comparison yielding an adjusted p-value of 0.214. Application of a global shrinkage factor of 0.8321 successfully adjusted the combined model calibration slope from an uncalibrated 0.794 to 0.965, alongside an overall Brier prediction error score of 0.196. Decision-curve analysis evaluating an enhanced monitoring strategy involving early six-week outpatient reassessments demonstrated crossing benefit curves across practical threshold probabilities, indicating that the combined imaging model did not uniformly surpass simpler screening strategies across all clinical risk tolerances.
Sensitivity analyses reinforced the consistency of the observed associations across several alternative clinical specifications, though findings remained contingent upon the exact outcome threshold applied. When defining unfavorable recovery as a twelve-month pain score of at least four points with less than thirty percent reduction, or as an absolute pain score of four or greater, both hyperintensity length and occult instability remained significantly associated with increased event odds. Conversely, when clinical failure was defined using an isolated pain score threshold of at least three points without requiring inadequate relative improvement, none of the target imaging parameters reached statistical significance. Linear multivariable regression examining continuous twelve-month pain scores confirmed that each five-millimeter increase in lesion length was associated with a 0.180-point increase in pain, while occult instability was associated with a 0.430-point increase. Methodological evaluations of sample size adequacy indicated that developing stable multivariable models under anticipated variance targets would require cohorts ranging from 438 to 597 patients, establishing that the current sample size of 325 individuals possessed limited statistical power to confirm subtle incremental discrimination gains. The authors concluded that while interspinous ligament lesion extent and dynamic instability offer valuable prognostic insights regarding persistent axial neck pain under routine care pathways, independent external multicenter validation and prospective impact trials remain essential before adopting these imaging markers into standardized clinical decision algorithms.
Subject of Research: Prognostic associations of cervical STIR hyperintensity length and occult instability with persistent neck pain
Article Title: Interspinous STIR hyperintensity and occult cervical instability in axial neck pain: a retrospective cohort study
Article References: Yang, J., Ma, J., Zhou, S., Batuer, C., & Liu, Y. (2026). Interspinous STIR hyperintensity and occult cervical instability in axial neck pain: a retrospective cohort study. Frontiers in Medicine, 13, Article 1975107. https://doi.org/10.3389/fmed.2026.1975107
Image Credits: AI Generated
DOI: 10.3389/fmed.2026.1975107
Keywords: cervical spondylosis, axial neck pain, STIR magnetic resonance imaging, interspinous ligament, occult cervical instability, flexion extension radiography, numerical rating scale, Neck Disability Index, multivariable logistic regression, prognostic modeling, internal validation, uniform shrinkage, area under the curve, decision curve analysis, dynamic translation, clinical prediction rules
News Source: Ophelia Keating. (November 19, 2026). Interspinous Ligament Hyperintensity Length and Occult Instability Linked to Persistent Neck Pain. Scienmag.



