Run iRECIST as a longitudinal state machine over RECIST 1.1 component data. Keep trial response assignment, clinical stability, protocol action and prognosis as separate outputs.
{
"categories": [
{
"outcome_code": "iCR",
"label": "immune_complete_response",
"target_lesions": "disappeared",
"non_target_lesions": "disappeared",
"new_lesions": "disappeared",
"lymph_nodes": "all_short_axis_below_10_mm",
"progression_override": "none"
},
{
"outcome_code": "iPR",
"label": "immune_partial_response",
"target_sum_change_from_baseline": "decrease_at_least_30_percent",
"overall_requirement": "no_iUPD_or_iCPD_override_after_target_non_target_and_new_lesion_synthesis"
},
{
"outcome_code": "iSD",
"label": "immune_stable_disease",
"state": "neither_iCR_nor_iPR_nor_progression_at_this_evaluable_timepoint"
},
{
"outcome_code": "iUPD",
"label": "immune_unconfirmed_progressive_disease",
"first_progression_trigger": "RECIST_1_1_progression_in_target_non_target_or_new_lesion_category",
"confirmation_status": "not_yet_confirmed"
},
{
"outcome_code": "iCPD",
"label": "immune_confirmed_progressive_disease",
"prerequisite": "prior_iUPD",
"confirmation_status": "qualifying_additional_progression_at_the_next_evaluable_assessment"
}
],
"applicability": {
"intended_use": "Standardized collection, time-point assignment and analysis of imaging response data in trials of immunotherapeutic agents when the protocol prospectively specifies iRECIST.",
"classification_unit": "one_trial_participant_at_one_protocol_defined_evaluable_timepoint_with_longitudinal_state_retained",
"core_boundary": "The 2017 authors describe iRECIST as recommendations for trial data handling, not validated response criteria. It must not be presented as a bedside treatment rule.",
"protocol_precondition": "Use only when the study protocol defines iRECIST acquisition, assessment, continuation-beyond-progression and analysis rules. Otherwise apply the response system actually specified by the protocol.",
"outside_scope": [
"routine_clinical_management_without_a_protocol",
"histology_specific_response_systems_when_the_protocol_requires_another_framework",
"metabolic_PET_response_without_an_explicit_protocol_mapping",
"diagnosing_pseudoprogression_from_one_scan",
"prognosis_or_survival_prediction_from_the_category_alone"
]
},
"baseline_RECIST_1_1_gate": {
"measurable_non_nodal_lesion": "long_axis_at_least_10_mm_on_a_scan_with_suitable_slice_thickness",
"measurable_lymph_node": "short_axis_at_least_15_mm",
"target_limit": "Select up to five target lesions total and up to two per organ, choosing representative lesions that can be measured reproducibly.",
"nontarget_rule": "All other known disease is recorded as non-target disease; pathologic nodes from 10 to below 15 mm short axis are non-target lesions.",
"baseline_sum": "Sum non-nodal long axes and nodal short axes for the fixed baseline target set.",
"nadir": "Use the smallest prior target-lesion sum on study when testing progression.",
"new_lesion_separation": "New lesions are not added to the baseline target sum. Track them in separate new-lesion target and non-target compartments."
},
"required_timepoint_inputs": {
"required": [
"protocol_and_treatment_context",
"baseline_target_lesion_identifiers_measurements_and_sum",
"baseline_non_target_disease",
"all_prior_evaluable_timepoint_states_and_target_sums",
"current_target_measurements",
"current_non_target_state",
"new_lesion_target_and_non_target_lists_with_first_seen_dates",
"scan_date_and_interval_from_any_prior_iUPD",
"technical_comparability_and_evaluability",
"clinical_stability_if_treatment_beyond_iUPD_is_being_considered"
],
"fixed_lesion_identity": "Never silently replace a baseline target lesion or merge a new lesion into the baseline target sum. Preserve lesion identifiers and explain any measurement exception.",
"full_patient_synthesis": "Assign one overall time-point response only after target, non-target and new-lesion compartments are each evaluated."
},
"first_progression_iUPD": {
"target_trigger": "At least 20 percent increase in the target sum from nadir and at least 5 mm absolute increase.",
"non_target_trigger": "Unequivocal progression of existing non-target disease.",
"new_lesion_trigger": "One or more new lesions, recorded and measured separately from baseline disease.",
"semantic_rule": "iUPD means progression is unconfirmed. It is neither proof of true progression nor proof of pseudoprogression.",
"unchanged_after_iUPD": "If the next evaluable assessment shows no qualifying further increase and no response/reset state, retain iUPD rather than manufacturing iCPD."
},
"confirmation_state_machine": {
"recommended_window": "Repeat imaging 4 to 8 weeks after iUPD.",
"early_scan_rule": "A scan before 4 weeks generally cannot confirm iCPD and should be disregarded for confirmation, except an out-of-schedule scan obtained for a new clinical symptom or sign that demonstrates an additional new lesion. Growth of an already known lesion alone is insufficient before 4 weeks.",
"delayed_scan_rule": "If scheduling or comorbidity prevents imaging by 8 weeks, the next evaluable scan may still confirm iCPD; retain the actual interval and protocol deviation.",
"target_confirmation": "At least 5 mm further increase in the target-lesion sum from the iUPD timepoint.",
"non_target_confirmation": "Any further increase in non-target disease after the iUPD non-target trigger can confirm.",
"new_target_confirmation": "At least 5 mm cumulative increase in the new-target-lesion sum from the iUPD reference; sequential increments are additive.",
"new_non_target_confirmation": "Any increase in existing new non-target disease, or an additional new lesion, can confirm.",
"cross_compartment_confirmation": "Confirmation need not arise from the same compartment as the original iUPD; RECIST progression in another compartment can establish iCPD.",
"response_after_iUPD": "If subsequent shrinkage meets iSD, iPR or iCR and progression is not confirmed, assign that response and reset the confirmation bar. A later qualifying progression is a new iUPD requiring confirmation.",
"less_than_5_mm_clarification": "For post-iUPD target or new-target sums, less than 5 mm change is treated as no confirming size increase, subject to the complete overall-state rules."
},
"clinical_stability_and_management_boundary": {
"continuation_condition": "The guideline permits protocol-defined treatment beyond iUPD only when the participant is clinically stable and the protocol/investigator allows it.",
"clinically_stable_requires": [
"no_worsening_of_performance_status",
"no_clinically_relevant_increase_in_disease_related_symptoms_such_as_pain_or_dyspnea",
"no_requirement_for_intensified_disease_related_symptom_management_including_more_analgesia_radiation_or_other_palliation"
],
"unstable_rule": "Clinical instability can require management change without waiting for radiologic confirmation. Imaging classification must not delay urgent care.",
"iCPD_rule": "iCPD is a confirmed trial response state, not an automatic stop order. Apply the protocol, clinical assessment, participant preferences and investigator judgment."
},
"new_lesion_tracking": {
"measurable_new_lesions": "Select up to five new target lesions in total and up to two per organ, using RECIST measurability principles; retain additional new disease as new non-target lesions.",
"first_seen_provenance": "Record every new lesion's site, identifier, first-seen timepoint, measurement status and subsequent trajectory.",
"equivocal_new_lesion": "If a possible new lesion is too small or equivocal, document uncertainty and confirm at follow-up rather than forcing progression from an unsubstantiated finding."
},
"missing_and_nonevaluable_assessments": {
"rule": "A missing or non-evaluable assessment is ignored for confirmation. The next evaluable assessment is the next assessment for the iUPD state machine.",
"prohibition": "Do not convert missing imaging into iCPD, iSD or a synthetic measurement.",
"partial_evaluability": "If one disease compartment is not evaluable and the overall state cannot be safely synthesized, return not evaluable with the known compartment findings and retained prior state."
},
"longitudinal_analysis": {
"iPFS_date": "When iCPD is confirmed, the progression event date used for immune progression-free survival is backdated to the initial iUPD date according to the prespecified analysis plan.",
"best_overall_response": "An iUPD does not permanently override a later iSD, iPR or iCR when progression is not confirmed.",
"trial_phase_boundary": "For phase III trials, the 2017 guideline recommended retaining RECIST 1.1 for primary efficacy analysis and using iRECIST as an exploratory framework unless the protocol and statistical plan specify otherwise.",
"no_prognostic_conversion": "Do not translate any iRECIST state into an individual survival probability or treatment-benefit claim without tumor-, therapy- and cohort-specific evidence."
},
"agent_output_contract": {
"always_return": [
"overall_timepoint_response",
"target_non_target_and_new_lesion_component_states",
"baseline_and_nadir_target_sums_with_units",
"prior_iUPD_date_and_confirmation_interval_when_applicable",
"new_lesion_identifiers_and_separate_sums",
"clinical_stability_source_or_unknown",
"technical_limitations_and_missing_assessments",
"protocol_name_and_response_framework"
],
"uncertainty_rule": "Expose the unresolved branch, missing input and next protocol-defined assessment; never guess clinical stability, lesion identity or confirmation."
},
"missing_input_behavior": [
"Without a RECIST-quality baseline and declared target set, do not assign iCR, iPR or target-based progression; return baseline-inadequate.",
"Without prior timepoints and nadir, do not compute target progression or confirm a prior iUPD.",
"Without the scan interval from iUPD, do not label size increase as timely iCPD confirmation.",
"Without clinical status, report clinical stability as unknown and do not recommend treatment beyond iUPD.",
"When a new lesion or measurement is equivocal, preserve uncertainty and request confirmation instead of forcing a category.",
"When the protocol does not specify iRECIST, do not silently substitute it for RECIST 1.1 or another required framework."
],
"supporting_sources": [
{
"role": "primary_guideline",
"citation": "Seymour et al. Lancet Oncol. 2017;18:e143-e152",
"doi": "10.1016/S1470-2045(17)30074-8",
"pmcid": "PMC5648544"
},
{
"role": "implementation_review",
"citation": "Persigehl et al. Cancer Imaging. 2020;20:2",
"doi": "10.1186/s40644-019-0281-x",
"pmcid": "PMC6942293"
},
{
"role": "current_official_clarifications",
"citation": "RECIST Working Group iRECIST page and FAQ",
"url": "https://recist.eortc.org/irecist/",
"reviewed_through": "2026-06-10"
}
],
"source_locator": "Seymour et al. 2017, PMC5648544, Tables 1-3 and sections on clinical stability, new lesions, time-point response and trial design; Persigehl et al. 2020, PMC6942293, Responses to therapy; RECIST Working Group iRECIST FAQ reviewed 2026-07-24 for early, delayed, missing and sequential-confirmation clarifications."
}