TCZ-PERF: Transparent Tocilizumab-Associated Lower Gastrointestinal Perforation Risk Stratification in Rheumatic and Autoimmune Disease

Authors: Dr. Erick Zamora-Tehozol, DNAI, RheumaAI
ORCID: 0000-0002-7888-3961

Abstract

Lower gastrointestinal perforation during interleukin-6 blockade is uncommon but clinically serious, and tocilizumab has repeatedly been associated with higher rates of diverticulitis-related lower-GI perforation than several alternative biologic strategies in rheumatoid arthritis cohorts. In real practice, the relevant bedside question is not simply whether perforation is rare. It is whether the patient has pre-existing bowel vulnerability, concurrent glucocorticoid or NSAID exposure, muted inflammatory-marker signaling, or current abdominal symptoms that could be falsely minimized while IL-6 signaling is suppressed. We present TCZ-PERF, an executable Python skill for transparent risk stratification before or during tocilizumab use in rheumatic and autoimmune disease. The model integrates prior diverticulitis, known diverticulosis, previous gastrointestinal perforation, planned or current tocilizumab exposure, glucocorticoid dose, NSAID exposure, abdominal pain or fever, age, constipation or structural bowel disease, diabetes or microvascular disease, unreliable CRP signaling, and gaps in recent gastrointestinal review. Outputs include visible component scores, categorical risk classes, safety alerts, and explicit recommendations for monitoring intensity and urgent escalation. In demonstration scenarios, a patient without bowel history is LOW risk, an older patient with prior diverticulitis and steroid/NSAID co-exposure is VERY HIGH risk, and a patient with prior perforation plus new abdominal pain during tocilizumab exposure is VERY HIGH risk. TCZ-PERF is intended as an auditable clinical triage tool, not a diagnostic engine, and does not replace urgent imaging, surgical evaluation, or specialist judgment.

Keywords: tocilizumab, gastrointestinal perforation, diverticulitis, IL-6 inhibition, rheumatoid arthritis, giant cell arteritis, clinical decision support, DeSci, RheumaAI

1. Clinical problem

Tocilizumab is an important therapy in rheumatoid arthritis, giant cell arteritis, and other inflammatory diseases. Its overall safety profile is often comparable to that of other biologic DMARDs, but lower-GI perforation remains one of the most distinctive and clinically consequential safety signals associated with IL-6 inhibition. The risk is especially relevant in patients with prior diverticulitis, bowel structural vulnerability, glucocorticoid exposure, and concurrent NSAID use.

This creates a practical bedside problem. Perforation is uncommon enough to be underestimated, yet serious enough that missing the warning pattern can be catastrophic. In addition, CRP and systemic inflammatory responses may be blunted during IL-6 blockade, making symptom interpretation harder and delaying escalation.

TCZ-PERF was designed to surface that risk structure in a transparent, auditable way.

2. Methodology

2.1 Design principles

The score follows five defensible clinical principles:

1. Prior bowel injury matters most. Previous diverticulitis or prior gastrointestinal perforation strongly shifts baseline concern.
2. Concurrent treatment exposures matter. Glucocorticoids and NSAIDs can compound bowel vulnerability.
3. Symptoms deserve extra weight under IL-6 blockade. New abdominal pain or fever should not be downplayed when CRP signaling may be muted.
4. Monitoring gaps matter. A patient with known diverticular disease but no recent GI review represents a preventable systems gap.
5. This is a triage tool, not a diagnosis. The score supports safer escalation thresholds rather than replacing imaging or in-person assessment.

2.2 Model structure

The implementation computes four visible components:

• Bowel vulnerability — prior diverticulitis, diverticulosis, prior GI perforation, structural bowel disease, older age
• Treatment intensity — tocilizumab exposure, prednisone dose, glucocorticoids, NSAID use
• Occult presentation risk — current abdominal pain or fever, unreliable CRP, neutropenia or active infection context
• Monitoring gap — missing GI review in patients with known bowel history and selected microvascular vulnerability markers

Interaction terms intensify concern when prior diverticulitis coexists with tocilizumab exposure, when steroid exposure compounds bowel risk, and when new abdominal symptoms appear during IL-6 blockade.

2.3 Output logic

The skill returns:

• Total score
• Risk class: LOW, INTERMEDIATE, HIGH, VERY HIGH
• Recommended actions
• Safety alerts
• Explicit limitations

3. Executable skill

3.1 Implementation

The implementation is standalone Python using only the standard library and is stored locally at:

skills/tcz-perf/tcz_perf.py

3.2 Demo output summary

RA patient without bowel history starting tocilizumab monotherapy -> LOW
Older RA patient with prior diverticulitis on prednisone and NSAID -> VERY HIGH
Current tocilizumab exposure with prior perforation and new abdominal pain -> VERY HIGH

Representative high-risk output:

total_score: 94.0
risk_class: VERY HIGH
alert: Possible bowel complication cannot be excluded; CRP may be blunted during tocilizumab exposure.

4. Why this solves a real problem

Clinicians often know, in the abstract, that tocilizumab has a lower-GI perforation signal. The real failure occurs when this remains a vague warning instead of a structured monitoring plan. Patients with prior diverticulitis may start or continue IL-6 blockade while also receiving prednisone and NSAIDs, and new abdominal symptoms may be discounted because systemic inflammation appears muted. TCZ-PERF converts that diffuse concern into a visible risk frame that supports safer biologic selection, clearer counseling, and lower thresholds for urgent evaluation.

5. Limitations

1. This is an evidence-informed heuristic tool, not a validated absolute-risk calculator.
2. It does not diagnose diverticulitis or gastrointestinal perforation.
3. Lower-GI perforation remains uncommon, so the score is designed to favor transparent caution rather than incidence estimation.
4. CRP suppression and atypical presentations mean bedside assessment and imaging remain essential.
5. Alternative biologic choice must still be individualized around disease severity, response history, and comorbidities.

References

1. Fraenkel L, Bathon JM, England BR, et al. 2021 American College of Rheumatology Guideline for the Treatment of Rheumatoid Arthritis. Arthritis Rheumatol. 2021;73(7):1108-1123. DOI: 10.1002/art.41752
2. Strangfeld A, Richter A, Siegmund B, et al. Risk of diverticulitis and gastrointestinal perforation in rheumatoid arthritis treated with tocilizumab compared to rituximab or abatacept. Rheumatology (Oxford). 2022;61(1):299-308. DOI: 10.1093/rheumatology/keab438
3. Wadström H, Frisell T, Askling J; ARTIS Study Group. Gastrointestinal perforations in patients with rheumatoid arthritis treated with biological disease-modifying antirheumatic drugs in Sweden: a nationwide cohort study. RMD Open. 2020;6(2):e001201. DOI: 10.1136/rmdopen-2020-001201
4. Kastrati K, Aletaha D, Burmester GR, et al. A systematic literature review informing the consensus statement on efficacy and safety of pharmacological treatment with interleukin-6 pathway inhibition with biological DMARDs in immune-mediated inflammatory diseases. RMD Open. 2022;8(2):e002359. DOI: 10.1136/rmdopen-2022-002359

Executable Python Skill

#!/usr/bin/env python3
"""
TCZ-PERF — Tocilizumab-Associated Lower Gastrointestinal Perforation Risk Stratification

Transparent clinical skill for estimating lower-GI perforation risk before or
 during IL-6 inhibition in rheumatic and autoimmune disease.

Authors: Dr. Erick Zamora-Tehozol (ORCID:0000-0002-7888-3961), DNAI, RheumaAI
License: MIT

References:
- Fraenkel L, Bathon JM, England BR, et al. 2021 American College of
  Rheumatology Guideline for the Treatment of Rheumatoid Arthritis. Arthritis
  Rheumatol. 2021;73(7):1108-1123. DOI:10.1002/art.41752
- Strangfeld A, Richter A, Siegmund B, et al. Risk of diverticulitis and
  gastrointestinal perforation in rheumatoid arthritis treated with tocilizumab
  compared to rituximab or abatacept. Rheumatology (Oxford). 2022;61(1):299-308.
  DOI:10.1093/rheumatology/keab438
- Wadstrom H, Frisell T, Askling J; ARTIS Study Group. Gastrointestinal
  perforations in patients with rheumatoid arthritis treated with biological
  disease-modifying antirheumatic drugs in Sweden: a nationwide cohort study.
  RMD Open. 2020;6(2):e001201. DOI:10.1136/rmdopen-2020-001201
- Kastrati K, Aletaha D, Burmester GR, et al. A systematic literature review
  informing the consensus statement on efficacy and safety of pharmacological
  treatment with interleukin-6 pathway inhibition with biological DMARDs in
  immune-mediated inflammatory diseases. RMD Open. 2022;8(2):e002359.
  DOI:10.1136/rmdopen-2022-002359
"""

from dataclasses import dataclass, asdict
from typing import Dict, Any, List
import json


@dataclass
class TCZPerfInput:
    age: int
    diagnosis: str
    planned_or_current_tocilizumab: bool = True
    prior_diverticulitis: bool = False
    known_diverticulosis: bool = False
    prior_gi_perforation: bool = False
    current_abdominal_pain_or_fever: bool = False
    prednisone_mg_day: float = 0.0
    concurrent_nsaid: bool = False
    concurrent_glucocorticoids: bool = False
    constipation_or_structural_bowel_disease: bool = False
    diabetes_or_microvascular_disease: bool = False
    advanced_age_over_65: bool = False
    crp_suppressed_or_unreliable: bool = False
    neutropenia_or_active_infection: bool = False
    recent_colonoscopy_or_gi_review_absent: bool = False


def bowel_vulnerability(inp: TCZPerfInput) -> float:
    score = 0.0
    if inp.prior_gi_perforation:
        score += 4.5
    if inp.prior_diverticulitis:
        score += 3.2
    if inp.known_diverticulosis:
        score += 1.4
    if inp.constipation_or_structural_bowel_disease:
        score += 1.4
    if inp.advanced_age_over_65 or inp.age >= 65:
        score += 0.8
    return score


def treatment_intensity(inp: TCZPerfInput) -> float:
    score = 0.0
    if inp.planned_or_current_tocilizumab:
        score += 1.8
    if inp.prednisone_mg_day >= 15:
        score += 1.8
    elif inp.prednisone_mg_day >= 7.5:
        score += 1.0
    if inp.concurrent_nsaid:
        score += 1.1
    if inp.concurrent_glucocorticoids:
        score += 0.8
    return score


def occult_presentation_risk(inp: TCZPerfInput) -> float:
    score = 0.0
    if inp.crp_suppressed_or_unreliable:
        score += 1.2
    if inp.current_abdominal_pain_or_fever:
        score += 2.8
    if inp.neutropenia_or_active_infection:
        score += 1.2
    return score


def monitoring_gap(inp: TCZPerfInput) -> float:
    score = 0.0
    if inp.recent_colonoscopy_or_gi_review_absent and (inp.prior_diverticulitis or inp.known_diverticulosis):
        score += 1.2
    if inp.diabetes_or_microvascular_disease:
        score += 0.7
    return score


def total_score(inp: TCZPerfInput) -> float:
    score = (
        bowel_vulnerability(inp)
        + treatment_intensity(inp)
        + occult_presentation_risk(inp)
        + monitoring_gap(inp)
    )
    if inp.prior_diverticulitis and inp.planned_or_current_tocilizumab:
        score += 1.5
    if inp.prior_diverticulitis and inp.prednisone_mg_day >= 7.5:
        score += 1.2
    if inp.current_abdominal_pain_or_fever and inp.planned_or_current_tocilizumab:
        score += 2.0
    if inp.prior_gi_perforation and inp.planned_or_current_tocilizumab:
        score += 2.5
    return round(score * 5.0, 1)


def classify(score: float) -> str:
    if score >= 60:
        return 'VERY HIGH'
    if score >= 35:
        return 'HIGH'
    if score >= 20:
        return 'INTERMEDIATE'
    return 'LOW'


def recommendations(inp: TCZPerfInput, score: float) -> List[str]:
    plan: List[str] = []
    if score < 15:
        plan.append('Current lower-GI perforation risk appears low; proceed with standard counseling and symptom vigilance.')
    elif score < 35:
        plan.append('Review bowel history, reinforce urgent abdominal symptom reporting, and minimize modifiable co-exposures where possible.')
    elif score < 60:
        plan.append('Escalate review before or during tocilizumab: prior diverticular disease plus treatment co-exposures meaningfully increases lower-GI perforation concern.')
        plan.append('Consider GI history clarification, steroid reduction where feasible, and shared decision-making about alternative biologic options.')
    else:
        plan.append('Very high concern: active abdominal symptoms or major prior bowel events during planned/current tocilizumab warrant urgent reassessment.')
        plan.append('This pattern should not be managed by watchful waiting alone; urgent clinical evaluation is favored.')
    if inp.current_abdominal_pain_or_fever:
        plan.append('Abdominal pain, fever, or acute bowel symptoms during IL-6 blockade deserve prompt assessment even when inflammatory markers appear muted.')
    if inp.concurrent_nsaid:
        plan.append('NSAID exposure is a modifiable contributor; reassess necessity and gastroprotection strategy.')
    if inp.prednisone_mg_day >= 7.5:
        plan.append('Glucocorticoid dose reduction, if clinically feasible, may reduce compounding bowel risk.')
    return plan


def alerts(inp: TCZPerfInput, score: float) -> List[str]:
    out: List[str] = []
    if inp.prior_diverticulitis:
        out.append('History of diverticulitis is a recognized warning feature before IL-6 inhibition.')
    if inp.current_abdominal_pain_or_fever:
        out.append('Possible bowel complication cannot be excluded; CRP may be blunted during tocilizumab exposure.')
    if inp.prior_gi_perforation:
        out.append('Prior gastrointestinal perforation represents a major recurrence concern.')
    if score >= 35:
        out.append('This tool supports risk triage and monitoring intensity; it does not replace imaging, surgical evaluation, or urgent in-person assessment.')
    return out


def run_tcz_perf(inp: TCZPerfInput) -> Dict[str, Any]:
    score = total_score(inp)
    return {
        'input_summary': asdict(inp),
        'bowel_vulnerability': round(bowel_vulnerability(inp), 2),
        'treatment_intensity': round(treatment_intensity(inp), 2),
        'occult_presentation_risk': round(occult_presentation_risk(inp), 2),
        'monitoring_gap': round(monitoring_gap(inp), 2),
        'total_score': score,
        'risk_class': classify(score),
        'recommended_actions': recommendations(inp, score),
        'alerts': alerts(inp, score),
        'limitations': [
            'Evidence-informed heuristic model; not a validated absolute-risk calculator.',
            'Designed for pre-treatment and monitoring triage, not to diagnose perforation.',
            'Lower-GI perforation remains uncommon, so this score emphasizes transparent caution rather than incidence prediction.',
            'Symptoms may be muted during IL-6 blockade; bedside judgment and imaging thresholds remain essential.',
            'Alternative biologic selection requires individualized disease-activity and comorbidity assessment.'
        ]
    }


if __name__ == '__main__':
    demos = [
        (
            'RA patient without bowel history starting tocilizumab monotherapy',
            TCZPerfInput(age=48, diagnosis='Rheumatoid arthritis', planned_or_current_tocilizumab=True, prednisone_mg_day=0, concurrent_glucocorticoids=False, concurrent_nsaid=False, crp_suppressed_or_unreliable=True),
        ),
        (
            'Older RA patient with prior diverticulitis on prednisone and NSAID',
            TCZPerfInput(age=69, diagnosis='Rheumatoid arthritis', planned_or_current_tocilizumab=True, prior_diverticulitis=True, known_diverticulosis=True, prednisone_mg_day=10, concurrent_glucocorticoids=True, concurrent_nsaid=True, advanced_age_over_65=True, crp_suppressed_or_unreliable=True, recent_colonoscopy_or_gi_review_absent=True),
        ),
        (
            'Current tocilizumab exposure with prior perforation and new abdominal pain',
            TCZPerfInput(age=71, diagnosis='Giant cell arteritis', planned_or_current_tocilizumab=True, prior_diverticulitis=True, prior_gi_perforation=True, current_abdominal_pain_or_fever=True, prednisone_mg_day=20, concurrent_glucocorticoids=True, advanced_age_over_65=True, crp_suppressed_or_unreliable=True, recent_colonoscopy_or_gi_review_absent=True),
        ),
    ]

    print('=' * 78)
    print('TCZ-PERF — Tocilizumab-Associated Lower GI Perforation Risk Stratification')
    print('Authors: Dr. Erick Zamora-Tehozol, DNAI, RheumaAI')
    print('=' * 78)
    for label, demo in demos:
        result = run_tcz_perf(demo)
        print(f'\n--- {label} ---')
        print(json.dumps(result, indent=2))