PeptideJournal
News13 min read

The Role of Peptides in Personalized Medicine

One-size-fits-all medicine is a poor fit for peptide therapeutics. A patient on semaglutide might lose 20% of their body weight while another loses 5% — same drug, same dose, vastly different outcomes.

One-size-fits-all medicine is a poor fit for peptide therapeutics. A patient on semaglutide might lose 20% of their body weight while another loses 5% — same drug, same dose, vastly different outcomes. A 2022 genome-wide study in The Lancet Diabetes & Endocrinology identified specific genetic variants that explain some of this variation, and machine learning models trained on proteomic data can now predict GLP-1 response with roughly 95% accuracy.

The era of prescribing peptides by body weight alone is ending. Pharmacogenomics, biomarker-guided dosing, and AI-driven drug design are converging to make peptide therapy more precise — and more effective — than it's ever been.

Table of Contents

Why Peptides Are Uniquely Suited to Precision Medicine

Peptide therapeutics occupy a molecular space between small-molecule drugs and large biologics like monoclonal antibodies. They're big enough to have specific receptor targets, but small enough to be modified, combined, and manufactured with relative flexibility. This makes them particularly amenable to personalized approaches.

Several characteristics set peptides apart. Their targets — specific receptors like GLP-1R, GIPR, somatostatin receptors, and GnRH receptors — are well-characterized at the genetic level, meaning we can identify variants that affect binding and signaling. Their metabolic pathways involve known enzymes (like DPP-4) where genetic variation has measurable effects. And their therapeutic windows are often wide enough to allow dose adjustments based on individual response without excessive toxicity risk.

The peptide therapeutics market, projected to reach roughly $100 billion by 2034, is increasingly moving toward stratified approaches. The question is no longer whether personalized peptide medicine will arrive — it's how fast the tools will mature.

Pharmacogenomics of GLP-1 Receptor Agonists

The most advanced pharmacogenomic work in the peptide field centers on GLP-1 receptor agonists. This makes sense: semaglutide, liraglutide, and their relatives are the highest-volume peptide drugs in the world, prescribed to millions of patients with widely varying responses.

The Foundational Genome-Wide Study

In 2022, Dawed et al. published the first comprehensive genome-wide analysis of GLP-1 receptor agonist pharmacogenomics in The Lancet Diabetes & Endocrinology. The study analyzed observational data and large randomized controlled trials to identify genetic variants associated with treatment response.

Two gene regions emerged as significant:

GLP1R (rs6923761, Gly168Ser): This common variant in the GLP-1 receptor gene was associated with blunted HbA1c reductions during GLP-1 receptor agonist treatment. The proposed mechanisms include reduced cell surface expression of GLP-1 receptors, altered intracellular calcium mobilization after GLP-1 stimulation, and reduced pancreatic GLP-1 receptor expression.

ARRB1 (beta-arrestin 1): Low-frequency variants in the gene encoding beta-arrestin 1 — a protein involved in receptor signaling and desensitization — were also associated with treatment response variation.

The clinical math is striking. Combining GLP1R and ARRB1 genotypes identified 4% of the population who had a 30% greater reduction in HbA1c than the 9% of the population with the worst response. Neither gene variant was associated with glycemic response to any non-GLP-1 receptor agonist drug, including placebo — strong evidence that these are drug-specific pharmacogenomic effects, not general metabolic traits.

What This Means for Prescribing

The implication is straightforward: when genotype data becomes routinely available at the point of prescribing, patients with certain ARRB1 variants might benefit from earlier initiation of GLP-1 receptor agonists, while patients with GLP1R variants associated with blunted response might be better served by alternative drug classes or higher doses from the start.

We're not there yet. Routine pharmacogenomic testing before peptide prescribing remains uncommon. But the data supporting it is accumulating rapidly.

The Non-Responder Problem

Large clinical trials and daily practice reveal a wide variability in response to GLP-1 drugs. Up to 30% of patients don't achieve even a minimum 5% weight loss compared to baseline — the threshold typically considered clinically meaningful. Real-world weight loss is "considerably lower" than clinical trial results, even for patients who adhere to therapy.

Understanding why some patients respond robustly while others barely respond at all is the central challenge of GLP-1 personalized medicine.

Tirzepatide and Dual-Agonist Pharmacogenomics

Tirzepatide, the dual GIP/GLP-1 receptor agonist, adds another layer of pharmacogenomic complexity. Because it activates two receptors instead of one, the potential sources of individual variation multiply.

A 2025 review in Pharmaceuticals mapped the current evidence for tirzepatide pharmacogenomics. Genetic variants in several gene families influence response:

  • GLP1R and GIPR polymorphisms affect receptor expression levels, signaling efficiency, and downstream metabolic pathways
  • TCF7L2 variants influence insulin secretion and beta-cell function
  • APOE variants modulate cardiovascular response
  • IL6 variants affect inflammatory and safety profiles
  • FTO and MC4R variants relate to appetite regulation and obesity susceptibility

The review concluded that integrating multi-omics data — combining genomic, proteomic, and metabolomic profiles — "enhances prediction accuracy and could enable precision stratification to optimize efficacy and minimize adverse effects."

This represents a shift from single-gene pharmacogenomics to systems-level patient profiling. For a dual-agonist like tirzepatide, single-gene analysis captures only part of the picture. The interaction between GIP receptor genetics, GLP-1 receptor genetics, metabolic gene variants, and inflammatory markers creates a complex response landscape that demands multi-dimensional analysis.

Beyond Genetics: Proteomic and Multi-Omics Responder Profiling

Genetics is only one piece of the personalization puzzle. Proteomic markers — the actual proteins circulating in a patient's blood — may be even more predictive of treatment response because they capture the current biological state, not just the inherited blueprint.

The 45-Protein Signature

In a study of 85 patients, Villikudathil et al. identified 45 plasma proteins that differed significantly between GLP-1 receptor agonist responders and non-responders. These proteins spanned extracellular matrix remodeling, inflammation, and vascular pathways. A machine learning model trained on this protein panel achieved approximately 95% predictive accuracy for treatment response, according to analysis cited in expert reviews of GLP-1 pharmacogenomics.

If validated in larger populations, this kind of proteomic panel could transform prescribing decisions. Instead of starting every patient on the same GLP-1 agonist dose and waiting months to assess response, a blood test could stratify patients into likely responders, partial responders, and non-responders before the first injection.

Sex-Based Response Differences

Multivariate regression analysis has revealed a consistent pattern: female sex predicts more pronounced HbA1c reductions and more effective body weight loss after GLP-1 receptor agonist therapy. This isn't pharmacogenomics in the traditional sense — it's a biological variable that affects drug response through hormonal, metabolic, and body composition differences.

Understanding these sex-based differences is part of the broader personalization picture. A truly precision approach to GLP-1 prescribing would integrate genetic variants, proteomic markers, sex, baseline metabolic status, and potentially gut microbiome composition into a composite response prediction.

Diabetes Subtyping

The pharmacogenomics field increasingly recognizes that type 2 diabetes itself is not a single disease. Data-driven cluster analysis has identified at least five diabetes subgroups with distinct phenotypes, complication risks, and genetic associations. Different subgroups may respond differently to GLP-1 agonists — a level of precision that current prescribing practices don't capture.

Biomarker-Guided Peptide Therapy in Oncology

Outside metabolic medicine, the most advanced application of personalized peptide therapy is in oncology. Peptide-drug conjugates (PDCs) and peptide-based radioligand therapies are fundamentally biomarker-guided — they only work when the target receptor is overexpressed on tumor cells.

Peptide-Drug Conjugates and Companion Diagnostics

PDCs represent a growing class of targeted cancer therapeutics. Unlike broad-spectrum chemotherapy, they combine a homing peptide (designed to bind a specific receptor) with a cytotoxic payload via a cleavable linker. The peptide delivers the drug directly to tumor cells expressing the target, sparing healthy tissue.

This approach is inherently personalized: it only works when the tumor expresses the target at sufficient levels. Current PDC targets include PSMA (prostate-specific membrane antigen) for prostate cancer, somatostatin receptors for neuroendocrine tumors, EphA2 for solid tumors, and TROP-2 for non-small cell lung cancer.

Lutathera (lutetium Lu-177 dotatate), the only FDA-approved PDC, treats somatostatin receptor-positive gastroenteropancreatic neuroendocrine tumors. Its use requires companion diagnostic imaging to confirm receptor expression — a working model of biomarker-guided peptide therapy.

Roughly 96 PDCs are now in development, with six in Phase III trials. The PDC market is projected to grow as companion diagnostic tools become more sophisticated, enabling better patient selection.

Theranostic Peptides: Diagnosis and Treatment in One

An emerging concept is theranostics — using the same peptide scaffold for both diagnostic imaging and therapeutic delivery. A peptide radiotracer can identify whether a patient's tumor expresses the target receptor (diagnosis), and if so, the same peptide can deliver a therapeutic radioisotope or cytotoxic payload (treatment).

PSMA-targeted theranostics in prostate cancer are the most developed example. Peptide-based PET imaging probes identify PSMA-positive tumors, and PSMA-targeted radioligand therapy delivers targeted radiation. This closed loop — diagnose with a peptide, treat with a peptide — is personalized medicine at the molecular level.

AI-Driven Peptide Design for Individual Patients

The intersection of artificial intelligence and peptide discovery is opening possibilities for personalization that weren't conceivable a decade ago.

De Novo Peptide Design

A study published in PMC described an approach that integrates deep learning-based protein design with functional screening. Researchers generated 10,000 de novo GLP-1 receptor agonists computationally, then screened them for stability, efficacy, and diversity. Sixty candidates satisfied all criteria in virtual functional screening — a hit rate that would have required years of traditional medicinal chemistry.

This capability doesn't yet translate to patient-specific drug design, but it points in that direction. If individual GLP-1 receptor variants affect drug binding and signaling, AI could theoretically design modified peptide agonists optimized for specific receptor genotypes.

Machine Learning for Response Prediction

The integration of pharmacogenomics and bioinformatics in anti-obesity drug research has produced machine learning models that stratify patients by predicted therapeutic response. These models use multi-omics datasets — combining genetic, proteomic, and metabolic data — to create response profiles that no single biomarker could achieve alone.

The computational paradigm shift from empirical screening to rational AI-guided design promises to overcome historical barriers in peptide therapeutics while unlocking new levels of precision in drug development and prescribing.

Individual Response Variation: What We Know So Far

Pulling the research together, here's what current evidence says about why patients respond differently to peptide therapeutics:

Genetic factors: Variants in GLP1R, ARRB1, GIPR, TCF7L2, FTO, MC4R, and other genes influence receptor binding, signaling efficiency, appetite regulation, and metabolic processing of peptide drugs.

Biological sex: Women tend to achieve greater HbA1c reductions and weight loss on GLP-1 receptor agonists than men, though the mechanisms aren't fully understood.

Disease subtype: Within type 2 diabetes, different disease subtypes (identified through cluster analysis) respond differently to incretin-based therapies.

Baseline metabolic status: Patients with higher baseline HbA1c levels tend to show greater absolute reductions, partly because there's more room for improvement.

Proteomic state: Circulating protein profiles reflecting inflammation, vascular health, and extracellular matrix remodeling predict GLP-1 response independently of genetic factors.

Adherence and pharmacokinetics: Real-world adherence is lower than clinical trial adherence, and individual variation in peptide absorption, distribution, and clearance affects drug exposure even at identical doses.

For a comparison of how different GLP-1 agonists perform, see our guide to semaglutide vs. tirzepatide.

The Future of Precision Peptide Dosing

Several developments are converging to make personalized peptide therapy a clinical reality:

Point-of-care genetic testing is becoming cheaper and faster. As pharmacogenomic panels expand to include peptide-relevant genes, prescribers will have genotype data available before writing the first prescription.

Proteomic blood panels that predict treatment response could become companion tests for GLP-1 prescribing, similar to how HER2 testing guides breast cancer treatment decisions.

Continuous glucose monitors and wearable metabolic sensors provide real-time data that can guide dose adjustments — a form of dynamic personalization that doesn't require genetic testing at all.

AI dosing algorithms that integrate genetic, proteomic, clinical, and wearable data into personalized dosing recommendations are in development. These systems could adjust peptide doses in real time based on individual response patterns.

Next-generation peptide drugs like orforglipron (an oral small-molecule GLP-1 agonist) and multi-agonists like retatrutide will create new pharmacogenomic questions — and new opportunities for personalized prescribing.

Barriers to Clinical Implementation

The science is advancing faster than the clinical infrastructure to use it. Several barriers remain:

Lack of prospective validation. Most pharmacogenomic associations for peptide drugs come from retrospective analyses. Large-scale, multi-ethnic, prospective trials that randomize treatment based on genotype are still needed.

Regulatory and reimbursement gaps. No regulatory framework requires or incentivizes pharmacogenomic testing before peptide prescribing. Insurance coverage for companion genetic testing is inconsistent.

Ethnic diversity in research. The foundational GLP-1 pharmacogenomics studies are heavily weighted toward European populations. Allele frequencies differ across ethnic groups, and response-associated variants identified in European cohorts may behave differently in African, Asian, or Latin American populations.

Clinical workflow integration. Even when genetic data exists, integrating it into prescribing decisions requires clinical decision support tools, physician education, and workflow changes that most healthcare systems haven't implemented.

Cost. Multi-omics profiling — combining genomics, proteomics, and metabolomics — remains expensive. For personalized peptide prescribing to scale, these costs need to drop to the level of routine blood work.

FAQ

Can genetic testing predict whether GLP-1 drugs will work for me?

Partially. Research has identified genetic variants in the GLP1R and ARRB1 genes that are associated with better or worse response to GLP-1 receptor agonists. However, routine clinical pharmacogenomic testing for GLP-1 response is not yet available or recommended by guidelines. The science is promising but still needs prospective validation before it's ready for widespread clinical use.

Why do some people lose 20% of their weight on semaglutide while others lose almost nothing?

Multiple factors contribute: genetic variants affecting GLP-1 receptor signaling, biological sex (women tend to respond better), baseline metabolic status, disease subtype, gut microbiome composition, adherence patterns, and individual variation in drug absorption and clearance. Up to 30% of patients don't achieve the 5% minimum weight loss threshold considered clinically meaningful.

Is personalized peptide dosing available now?

Not in the pharmacogenomic sense — no clinical guidelines currently recommend genetic testing before prescribing peptide therapeutics. However, clinicians do personalize dosing based on individual response, titrating doses up or down based on efficacy and side effects. This trial-and-error approach is standard practice but far less efficient than biomarker-guided prescribing could be.

How are peptides used in personalized cancer treatment?

Peptide-drug conjugates (PDCs) and radioligand therapies are inherently personalized because they target specific receptors overexpressed on tumors. Patients must be tested for receptor expression before treatment — a form of biomarker-guided precision therapy. Lutathera for neuroendocrine tumors is the primary approved example, with roughly 96 more PDCs in development.

What role does AI play in personalized peptide medicine?

AI contributes in two ways. First, machine learning models trained on multi-omics data can predict individual treatment response more accurately than any single biomarker. Second, AI-driven peptide design enables rapid creation of new peptide candidates optimized for specific receptor variants or patient profiles. Both applications are still largely in the research phase but advancing rapidly.

The Bottom Line

Personalized peptide medicine is no longer theoretical. Specific genetic variants predict GLP-1 response. Proteomic panels achieve 95% accuracy in distinguishing responders from non-responders. Peptide-drug conjugates in oncology already require biomarker testing for patient selection. And AI is designing peptide candidates at speeds that would have seemed impossible five years ago.

The gap is between research and clinical practice. The pharmacogenomic data exists; the clinical infrastructure to use it doesn't — not yet. Routine pre-prescribing genetic panels, proteomic blood tests for response prediction, and AI-integrated dosing algorithms are all technically feasible. What's needed is prospective validation, regulatory frameworks, reimbursement pathways, and clinical workflow integration.

For patients today, the practical takeaway is simpler: if you're not responding to a peptide therapy as expected, the cause may be biological, not behavioral. The science of individual variation in peptide response is real, it's measurable, and it's rapidly becoming actionable.

References

  1. Dawed AY, et al. "Pharmacogenomics of GLP-1 receptor agonists: a genome-wide analysis of observational data and large randomised controlled trials." The Lancet Diabetes & Endocrinology. 2023;11(1):55-66. PubMed

  2. Nasykhova YA, et al. "Pharmacogenomics of Tirzepatide: Genomic Insights into Dual GIP/GLP-1 Agonist Response in Type 2 Diabetes and Atherosclerosis." Pharmaceuticals. 2025;18(9):1261. PMC

  3. Rial SA, et al. "Predicting treatment response to GLP-1 receptor agonists: still tossing the coin or doing better?" Expert Opinion on Pharmacotherapy. 2025. Full Text

  4. Armstrong E, et al. "Peptide-Drug Conjugates: A New Hope for Cancer." Journal of Peptide Science. 2025. PMC

  5. Drucker DJ. "The expanding landscape of GLP-1 medicines." Nature Medicine. 2025. Nature

  6. Yang W, et al. "Emerging Frontiers in GLP-1 Therapeutics: A Comprehensive Evidence Base (2025)." Pharmaceutics. 2025;17(8):1036. PMC

  7. Ahlqvist E, et al. "Pharmacogenetics of novel glucose-lowering drugs." Diabetologia. 2021;64:1159-1169. Springer

  8. Ghosal S, et al. "Pharmacogenomics of Anti-Obesity Drugs: A Bioinformatics Approach." Research Square. 2025. Preprint

  9. Zhang R, et al. "AI-Driven De Novo Design of Ultra Long-Acting GLP-1 Receptor Agonists." Advanced Science. 2025. PMC

  10. Therapeutic Peptides: Recent Advances in Discovery, Synthesis, and Clinical Translation. Molecules. 2025. PMC

  11. Innovative Peptide Therapeutics in the Pipeline: Transforming Cancer Detection and Treatment. Cancers. 2025. PMC