Background

Stochastic portfolio theory (SPT), developed by Fernholz (2002) for equity markets, provides a mathematical framework for analyzing the behavior of diverse portfolios under stochastic dynamics without requiring parametric assumptions about asset returns. We propose applying SPT concepts — particularly the excess growth rate generated by portfolio diversity — to the problem of multi-drug regimen optimization in systemic lupus erythematosus (SLE).

Hypothesis

We hypothesize that treating concurrent immunosuppressive agents as "assets" in a therapeutic portfolio, where each drug's efficacy trajectory follows a semimartingale process conditioned on patient-specific pharmacogenomic covariates (NAT2, CYP2C19, TPMT, UGT1A9), a diversity-weighted therapeutic portfolio — one that maintains sufficient heterogeneity across mechanistic pathways (complement inhibition, B-cell depletion, calcineurin inhibition, antimetabolites) — will generate an excess therapeutic return analogous to the excess growth rate in SPT, resulting in >20% greater cumulative SLEDAI reduction over 24 months compared to sequential monotherapy switching.

Mathematical Framework

Let $X_i(t)$ represent the efficacy process of drug $i$ at time $t$, modeled as:

$$dX_i(t) = X_i(t)[\gamma_i(\mathbf{G}) dt + \sigma_i(\mathbf{G}) dW_i(t)]$$

where $\gamma_i(\mathbf{G})$ is the pharmacogenomically-conditioned drift (expected efficacy rate) and $\sigma_i(\mathbf{G})$ captures stochastic variability dependent on genotype vector $\mathbf{G}$. The portfolio diversity measure $D(\boldsymbol{\mu}(t)) = -\sum_{i} \mu_i(t) \log \mu_i(t)$ (where $\mu_i(t)$ is the relative therapeutic weight of drug $i$) generates excess growth rate $\gamma^* \geq \frac{1}{2}\sum_i \mu_i \sigma_i^2 (1 - \mu_i)$.

This implies that even when individual drug efficacies are comparable, the volatility harvesting effect from maintaining mechanistic diversity provides a systematic therapeutic advantage — the portfolio outperforms its average constituent.

Testable Predictions

1. Primary: In a simulated cohort of 500 SLE patients with known pharmacogenomic profiles, diversity-weighted multi-drug portfolios achieve >20% greater cumulative SLEDAI area-under-curve reduction versus sequential monotherapy over 24 months.
2. Secondary: The excess therapeutic return correlates positively with inter-drug efficacy volatility — patients with high pharmacogenomic variability in drug response benefit more from diversified regimens.
3. Tertiary: The optimal portfolio diversity level (Shannon entropy of drug weights) follows a concave relationship with clinical benefit, with diminishing returns beyond 3–4 mechanistically distinct agents due to interaction toxicity.
4. Validation: Retrospective analysis of BILAG/BLISS trial data comparing patients on combination therapy versus sequential monotherapy should reveal a diversity premium consistent with SPT predictions.

Limitations

• Drug interactions introduce non-independent covariance structures not present in financial portfolios; toxicity cross-terms may violate SPT's diversity benefit under certain regimens.
• The analogy assumes continuous rebalancing, whereas clinical dose adjustments occur at discrete intervals (weeks to months), introducing path-dependent tracking error.
• Pharmacogenomic drift parameters require longitudinal estimation; single-timepoint genotyping may miss epigenetic or microbiome-mediated shifts in drug metabolism.
• SPT's theoretical guarantees rely on non-degeneracy conditions that may not hold during severe flares (immunological regime changes).
• Retrospective validation from existing trials is confounded by indication bias — sicker patients receive more drugs.

Clinical Significance

If validated, SPT-based therapeutic portfolio optimization could formalize the empirical observation that combination immunosuppression often outperforms sequential monotherapy beyond simple additive effects. It provides a principled, pharmacogenomically-informed framework for determining how many and which drugs to combine, with patient-specific diversity targets. This could reduce the current trial-and-error approach to combination therapy in SLE and provide actuarial-grade quantification of expected therapeutic benefit from regimen diversification.

RheumaAI Research • rheumai.xyz • DeSci Rheumatology