Classify endplate-adjacent marrow per level from sequence-aware T1/T2 evidence, preserve mixed patterns and dangerous mimics, and never turn a Modic type into pain causality, infection exclusion or an autonomous treatment instruction.
{
"categories": [
{
"outcome_code": "1",
"conventional_T1": "low_relative_to_normal_vertebral_marrow",
"conventional_non_fat_suppressed_T2": "high",
"tissue_correlate": "endplate_disruption_and_fissuring_with_fibrovascular_marrow_replacement_and_increased_water",
"shorthand": "fibrovascular_or_edematous_pattern"
},
{
"outcome_code": "2",
"conventional_T1": "high",
"conventional_non_fat_suppressed_T2": "isointense_to_slightly_high",
"tissue_correlate": "fatty_yellow_marrow_replacement",
"shorthand": "fatty_pattern"
},
{
"outcome_code": "3",
"conventional_T1": "low",
"conventional_non_fat_suppressed_T2": "low",
"tissue_correlate": "subchondral_bone_sclerosis",
"shorthand": "sclerotic_pattern"
}
],
"applicability": {
"intended_use": "Describe vertebral endplate-adjacent bone-marrow signal associated with a degenerative disc at each named spinal level.",
"original_population": "The original classification was developed in lumbar degenerative disc disease; application in the cervical or thoracic spine should retain the region and be labeled as extrapolation rather than silent original validation.",
"unit_of_analysis": "A disc/endplate level, with the cranial and caudal vertebral endplates, side, anterior-posterior distribution and extent retained rather than a single patient-level type.",
"required_sequences": [
"non_fat_suppressed_T1_weighted_MRI",
"non_fat_suppressed_T2_weighted_MRI_or_sequence_context_that_allows_equivalent_interpretation"
],
"outside_scope": [
"disc_degeneration_grade_use_Pfirrmann_separately",
"disc_herniation_or_annular_fissure",
"canal_or_foraminal_stenosis",
"proof_of_discitis_osteomyelitis",
"identification_of_a_pain_generator",
"treatment_selection"
]
},
"technical_and_reporting_gate": {
"reference": "Compare endplate-adjacent signal with normal vertebral marrow on the same examination and interpret it in the patient's age and marrow context.",
"sequence_dependency": "Field strength, echo time, fat suppression, spatial resolution and other acquisition parameters change conspicuity and apparent extent; record the sequences used when serially measuring or researching Modic change.",
"fat_suppression_trap": "Fat in type 2 becomes dark on fat-suppressed T2/STIR, so a low signal on those sequences is not sufficient for type 3. Type 3 requires low signal on conventional T1 and non-fat-suppressed T2 plus a sclerotic correlate when available.",
"CT_radiograph_role": "CT or radiographs may corroborate sclerosis in type 3 but cannot independently assign the MRI signal classification.",
"comparison_rule": "Serial comparison should use technically comparable sequences; otherwise qualify apparent change in type or extent as technically limited."
},
"category_algorithm": [
{
"step": 1,
"pattern": "low_T1_and_high_non_fat_suppressed_T2_adjacent_to_degenerative_endplate",
"output": "1"
},
{
"step": 2,
"pattern": "high_T1_and_iso_to_slightly_high_non_fat_suppressed_T2",
"output": "2"
},
{
"step": 3,
"pattern": "low_T1_and_low_non_fat_suppressed_T2_with_sclerotic_morphology",
"output": "3"
},
{
"step": 4,
"pattern": "more_than_one_pattern_within_the_same_level_or_endplate",
"output": "mixed_type_with_each_component_retained"
},
{
"step": 5,
"pattern": "signal_does_not_match_or_required_sequences_are_missing",
"output": "indeterminate_or_non_Modic_differential"
}
],
"type_and_mixed_pattern_rules": {
"type_1": "Low T1/high T2 reflects a water-rich fibrovascular endplate phenotype. The colloquial word active must not be converted into proof of painful inflammation or infection.",
"type_2": "High T1 with iso/slightly high T2 reflects fatty marrow replacement; use the T1 fat signal and non-fat-suppressed sequence context to avoid misclassification.",
"type_3": "Low T1/low T2 reflects sclerosis and was detailed in the companion 1988 imaging paper. It is less common and should not be assigned from low STIR signal alone.",
"mixed_patterns": "Mixed I/II and II/III patterns occur. Report both components and their spatial distribution; do not force the dominant area into a pure type or create an unsupported arithmetic score.",
"nonordinal_rule": "Types are tissue-signal phenotypes, not three validated steps of symptom severity, treatment urgency or inevitable temporal progression."
},
"minimum_structured_report": {
"anatomy": [
"disc_level",
"cranial_endplate",
"caudal_endplate",
"laterality_or_centrality",
"anterior_middle_posterior_distribution"
],
"phenotype": [
"pure_or_mixed_type",
"component_types",
"qualitative_or_contoured_extent",
"dominant_component_only_if_explicitly_labeled"
],
"companions": [
"disc_degeneration_and_height",
"endplate_defect_or_Schmorl_node",
"erosion_or_destruction",
"sclerosis",
"enhancement_if_contrast_given",
"paraspinal_or_epidural_abnormality",
"comparison_change"
],
"reproducibility_context": [
"field_strength",
"T1_parameters",
"T2_parameters",
"fat_suppression",
"slice_thickness_and_orientation"
]
},
"type_1_differential_and_safety_gate": {
"mimics": [
"infectious_spondylodiscitis",
"acute_Schmorl_node",
"acute_or_subacute_fracture",
"inflammatory_spondyloarthropathy",
"tumor_or_treatment_related_marrow_change"
],
"infection_supporting_features": [
"progressive_endplate_erosion_or_destruction",
"fluid_like_or_markedly_high_disc_T2_signal",
"disc_and_endplate_enhancement_in_context",
"paraspinal_or_epidural_phlegmon_or_abscess",
"systemic_or_laboratory_evidence_of_infection"
],
"enhancement_boundary": "Endplate and disc enhancement can occur with degeneration and infection; enhancement alone does not settle the diagnosis.",
"diffusion_claw": "On appropriate DWI, a paired well-marginated linear claw at the normal-abnormal marrow interface supports degeneration, while diffuse amorphous DWI signal raises infection concern. This is supportive evidence from a retrospective cohort, not part of the original Modic category or an absolute exclusion test.",
"action_rule": "If infection, fracture or tumor is plausible, surface that differential and recommend context-appropriate diagnostic escalation independently of the Modic label; never reassure from type 1 alone."
},
"symptoms_risk_and_management_context": {
"pain_association": "A 2018 systematic review of 31 studies found inconsistent associations between Modic change and low-back-pain or activity-limitation outcomes, with substantial heterogeneity and high risk of bias.",
"pain_generator_rule": "No Modic type proves that the indexed level generates pain, and absence of Modic change does not exclude clinically important discogenic or other spinal pain.",
"prognosis_rule": "Type and longitudinal conversion do not provide a calibrated individual forecast of pain, disability, degeneration rate, fracture, cancer or surgical outcome.",
"treatment_rule": "The classification does not choose observation, physical therapy, injection, surgery, ablation or medication. Integrate symptoms, examination, other imaging findings, duration, prior care and patient goals.",
"antibiotic_evidence": "The 2019 AIM randomized trial found no clinically important one-year benefit from three months of amoxicillin in its selected chronic-low-back-pain population. A 2026 Cochrane review found only low-certainty slight-to-small short-term pain benefit and small-to-moderate disability benefit in selected patients with type 1 change plus disc herniation, with very uncertain harms; a Modic image is therefore not a standalone antibiotic indication.",
"stewardship_rule": "Do not recommend or prescribe antibiotics from the type alone. Suspected spinal infection follows an infection workup; experimental or selected chronic-pain antibiotic evidence requires specialist clinical judgment and antimicrobial stewardship."
},
"longitudinal_rules": {
"preserve_trajectory": "Record prior and current component types and extents at each level; mixed change can evolve in more than one direction and should not be reduced to a presumed fixed sequence 1-to-2-to-3.",
"technical_change": "Before calling conversion or enlargement, check sequence comparability and fat suppression.",
"clinical_change": "New fever, bacteremia risk, severe rest/night pain, neurologic deficit or rapidly destructive morphology overrides a routine degenerative follow-up interpretation."
},
"agent_output_contract": [
"each_disc_level_and_each_adjacent_endplate",
"pure_mixed_or_indeterminate_Modic_phenotype",
"T1_and_non_fat_suppressed_T2_signal_evidence",
"spatial_extent_and_distribution",
"sequence_and_comparison_adequacy",
"disc_endplate_paraspinal_epidural_and_sclerotic_companions",
"infection_fracture_inflammatory_and_tumor_differential_when_relevant",
"no_pain_generator_prognosis_or_treatment_inference_from_type_alone"
],
"missing_input_behavior": [
"If conventional T1 is unavailable, do not assign type 1 versus type 2 from fluid-sensitive imaging alone.",
"If only fat-suppressed T2/STIR is available, do not assign type 2 versus type 3 from low signal; return sequence-limited.",
"If two tissue patterns coexist, return a mixed type and map each component instead of selecting one code.",
"If destructive endplate, disc, paraspinal or systemic features raise infection, return Modic-versus-infection uncertainty and prioritize the safety differential.",
"If symptoms and examination are absent, report morphology only and withhold pain causality and management conclusions."
],
"supporting_sources": [
{
"role": "primary_types_1_and_2",
"citation": "Modic et al. Radiology. 1988;166:193-199",
"doi": "10.1148/radiology.166.1.3336678",
"pmid": "3336678"
},
{
"role": "type_3_and_imaging_context",
"citation": "Modic et al. Radiology. 1988;168:177-186",
"doi": "10.1148/radiology.168.1.3289089",
"pmid": "3289089"
},
{
"role": "reporting_recommendations",
"citation": "Fields et al. Eur Spine J. 2019;28:2266-2274",
"doi": "10.1007/s00586-019-06119-6",
"pmcid": "PMC7205555"
},
{
"role": "pain_association_limit",
"citation": "Herlin et al. PLoS One. 2018;13:e0200677",
"doi": "10.1371/journal.pone.0200677",
"pmcid": "PMC6070210"
},
{
"role": "infection_differential",
"citation": "Patel et al. AJNR. 2014;35:1647-1652",
"doi": "10.3174/ajnr.A3948",
"pmcid": "PMC7964436"
},
{
"role": "antibiotic_randomized_trial",
"citation": "Bråten et al. BMJ. 2019;367:l5654",
"doi": "10.1136/bmj.l5654",
"pmcid": "PMC6812614"
},
{
"role": "current_antibiotic_synthesis",
"citation": "Liu et al. Cochrane Database Syst Rev. 2026;4:CD014221",
"doi": "10.1002/14651858.CD014221.pub2",
"pmid": "41944232"
}
],
"source_locator": "Modic et al. 1988, DOI 10.1148/radiology.166.1.3336678, abstract and histologic correlation for types 1-2; Modic et al. 1988, DOI 10.1148/radiology.168.1.3289089 for type 3; Fields et al. 2019, PMC7205555, Recommendations and Tables for acquisition, measurement, mixed types and reporting; Herlin et al. 2018, PMC6070210, Results/Conclusions for inconsistent pain associations; Patel et al. 2014, PMC7964436, Methods/Results for the DWI claw sign; Bråten et al. 2019, PMC6812614, Results/Conclusions; Liu et al. 2026 Cochrane review, DOI 10.1002/14651858.CD014221.pub2, Authors' conclusions."
}