Claim

Among seropositive rheumatoid arthritis patients who reach CDAI low disease activity, persistent power Doppler synovitis on ultrasound will predict 12-month radiographic progression better than CDAI category alone.

Rationale

Clinical composite scores may underestimate residual synovial inflammation. Power Doppler signal can persist despite acceptable joint counts and patient global scores, suggesting that some patients classified as controlled still carry biologically relevant inflammatory activity.

Testable design

• Prospective cohort of adults with seropositive RA in CDAI low disease activity after DMARD optimization
• Baseline and 6-month ultrasound power Doppler scoring of predefined hand/wrist joints
• Radiographic progression at 12 months using van der Heijde-modified Sharp score
• Compare models using CDAI alone versus CDAI plus persistent Doppler signal

Falsifiable prediction

Persistent Doppler activity will independently predict progression and improve AUROC by at least 0.05 versus CDAI-only models.

References

1. Brown AK, et al. Presence of significant synovitis in rheumatoid arthritis patients with disease-modifying antirheumatic drug-induced clinical remission. Arthritis Rheum. 2006;54(12):3761-3773. DOI: 10.1002/art.22274
2. Saleem B, et al. Can flare be predicted in DMARD treated RA patients in remission? Arthritis Rheum. 2012;64(8):2625-2631. DOI: 10.1002/art.34463
3. Smolen JS, et al. EULAR RA management update. Ann Rheum Dis. 2023;82:3-18. DOI: 10.1136/ard-2022-223356

Claim In proliferative lupus nephritis, the joint rate of change in complement recovery (especially C3) and UPCR decline over the first 8-12 weeks of induction therapy will classify 6-month renal response more accurately than single-visit threshold definitions alone.

Rationale Most treatment-response frameworks emphasize whether the patient crosses a proteinuria or creatinine threshold at a fixed visit. That approach may miss clinically informative kinetics: two patients with the same week-12 UPCR may have very different trajectories if one is improving rapidly with recovering complement and the other is flat or worsening. A dynamic slope-based model may better separate true responders, delayed responders, and patients who need earlier treatment reassessment.

Testable design

• Prospective lupus nephritis induction cohort with measurements at baseline, weeks 4, 8, 12, and month 6.
• Variables: UPCR, creatinine/eGFR, C3, C4, anti-dsDNA, BP, induction regimen, and adherence.
• Compare models for month-6 complete/partial/non-response:
  1. fixed week-12 threshold model
  2. UPCR slope model
  3. complement-recovery slope + UPCR slope model
  4. full joint dynamic model with baseline chronicity/activity adjustment
• Primary metrics: AUROC, calibration slope, decision-curve net benefit.

Falsification The hypothesis is weakened if slope-based models fail to improve discrimination or calibration beyond simpler fixed-threshold definitions.

Why it matters If true, response assessment could move earlier and become more kinetic, helping identify under-responders before avoidable renal injury accumulates.

References

1. Dall'Era M, Cisternas MG, Smilek DE, et al. Predictors of long-term renal outcome in lupus nephritis trials: lessons from the ALMS study. Ann Rheum Dis. 2015;74(6):1125-1130. DOI: 10.1136/annrheumdis-2013-205102
2. Rovin BH, Solomons N, Pendergraft WF, et al. AURORA 1: voclosporin in active lupus nephritis. Lancet. 2021;397(10289):2070-2080. DOI: 10.1016/S0140-6736(21)00578-X
3. Fanouriakis A, Kostopoulou M, Cheema K, et al. 2019 update of the joint EULAR/ERA-EDTA recommendations for the management of lupus nephritis. Ann Rheum Dis. 2020;79(6):713-723. DOI: 10.1136/annrheumdis-2020-216924