Type 2 diabetes mellitus (T2DM) and obesity are closely interconnected conditions that reinforce one another through insulin resistance (IR), chronic low-grade inflammation, adipose tissue dysfunction, ectopic fat accumulation, and broader cardiometabolic dysregulation [1, 2]. The term “diabesity”, commonly attributed to Ethan A. H. Sims and colleagues in the 1970s, was introduced to emphasize the close metabolic relationship between excess adiposity and diabetes-related dysglycemia [1]. In this review, the term is used descriptively to refer to the coexistence of obesity and T2DM and should not be interpreted as a separate disease entity or independent clinical diagnosis. Coexisting T2DM and obesity are associated with a higher burden of cardiovascular (CV), kidney, hepatic, and functional complications, with important consequences for quality of life (QoL), mortality, and healthcare expenditure [3–5]. Their increasing prevalence reflects changes in lifestyle, nutrition, urbanization, population aging, and environmental exposures as well as substantial biological heterogeneity [6, 7]. Excess adiposity, particularly visceral and ectopic fat accumulation, promotes IR, inflammation, adipokine dysregulation, hepatic steatosis, and progressive impairment of glucose homeostasis. Obesity is therefore both a major risk factor for T2DM and an important modifier of its clinical course, complications, and response to pharmacological treatment [6–10].

The global burden of obesity has increased markedly over recent decades. According to the World Health Organization (WHO), approximately 2.5 billion adults were overweight in 2022, including about 890 million adults with obesity, corresponding to 43% and 16% of the adult population, respectively [6]. Large pooled analyses have shown that adult obesity prevalence more than doubled between 1990 and 2022, while obesity among children and adolescents increased even more rapidly [8]. Projections from the World Obesity Federation (WOF) suggest that the number of adults living with obesity may rise to 1.53 billion by 2035 if current trends continue [7]. This trajectory has major clinical and economic implications, with the annual global economic impact of overweight and obesity projected to exceed USD 4 trillion by 2035 [9].

In parallel, diabetes mellitus (DM) has become a major global health challenge. In 2021, approximately 536.6 million adults aged 20–79 years were estimated to be living with DM worldwide, corresponding to a prevalence of 10.5% [11], and this number was projected to increase to 783.2 million by 2045. More recent International Diabetes Federation estimates indicate that 589 million adults are living with DM and that this number may rise to approximately 853 million by 2050 [12]. Because more than 90% of DM cases are T2DM, the increasing burden of T2DM is closely linked to the global expansion of obesity and related cardiometabolic risk factors [12].

The clinical consequences of coexisting T2DM and obesity extend beyond hyperglycemia and excess body weight [13–15]. Both conditions increase the risk of cardiovascular disease (CVD), chronic kidney disease (CKD), metabolic dysfunction-associated steatotic liver disease (MASLD), obstructive sleep apnea (OSA), heart failure (HF), osteoarthritis, and several obesity-related cancers [10, 13–15]. Importantly, patients with T2DM and obesity do not represent a homogeneous population [16]. Individuals differ in the relative contribution of IR, beta-cell dysfunction, visceral adiposity, liver fat accumulation, inflammation, genetic susceptibility, gut microbiome composition, comorbidity burden, and environmental exposures. This heterogeneity contributes to substantial interindividual variability in therapeutic efficacy, tolerability, risk-factor modification, and treatment durability [10, 13–15].

Genome-wide association studies have identified multiple loci linked to T2DM and CVD [17], but genetic signals explain only part of the observed disease heterogeneity. Contemporary treatment therefore increasingly extends beyond glucose lowering to address body weight, cardiometabolic risk factors, liver disease, and, where supported by dedicated outcome trials, CV and kidney outcomes. These outcome domains should be interpreted separately: reductions in body weight or HbA1c and improvements in blood pressure (BP), lipid profile, liver fat, or albuminuria are clinically relevant but should not automatically be considered evidence of CV or kidney event reduction. Although incretin-based therapies have expanded treatment options, selection among available and emerging therapies remains largely empirical, and validated predictive biomarkers are limited [16, 17].

Previous reviews have examined precision medicine in obesity, multi-omics approaches to obesity management, heterogeneity-based precision medicine in T2DM, and the emerging field of pharmacomulti-omics in diabetes [18–21]. The distinctive contribution of the present review is not the introduction of multi-omics as a new concept, but the integration of the rapidly evolving anti-obesity pharmacotherapy pipeline with an explicit translational hierarchy of evidence in the specific context of coexisting T2DM and obesity. We distinguish population-level clinical efficacy from mechanistic rationale, candidate response biomarkers, exploratory associations, and hypothetical future treatment-selection strategies. Particular emphasis is placed on the gap between biomarker discovery and clinically actionable prescribing, including the validation requirements for treatment combinations, sequencing, body-composition outcomes, and organ-specific risk.

Accordingly, the aim of this narrative review is to summarize contemporary and emerging anti-obesity pharmacotherapies and to critically evaluate how, and at what current level of evidence, multi-omics could contribute to precision pharmacotherapy in patients with coexisting T2DM and obesity.

This narrative review was developed through a structured search of peer-reviewed literature complemented by searches of clinical trial registries and official regulatory or sponsor sources. PubMed/MEDLINE was used as the primary bibliographic database. ClinicalTrials.gov was searched to identify ongoing, recently completed, and late-stage development programs relevant to obesity pharmacotherapy and T2DM. Official regulatory documents and company communications were consulted when recent information on regulatory status, development programs, or preliminary results was not yet available in peer-reviewed publications. The search covered literature available through August 2026; no predefined lower publication-date limit was applied.

Search concepts included combinations of terms related to T2DM, obesity, diabesity, precision medicine, multi-omics, genomics, epigenomics, transcriptomics, proteomics, metabolomics, lipidomics, microbiome, pharmacometabolomics, anti-obesity medications, glucagon-like peptide-1 (GLP-1) receptor agonists, glucose-dependent insulinotropic polypeptide (GIP), glucagon receptor agonism, dual receptor agonists, triple receptor agonists, amylin, body composition, treatment response, biomarkers, and the names of individual approved or investigational agents discussed in this review. Priority was given to peer-reviewed randomized controlled trials, particularly phase 2 and phase 3 studies, mechanistic studies, consensus or guideline documents, and human studies evaluating omics-based stratification, biomarkers, pharmacological response, or precision medicine in T2DM and obesity. Studies were considered when they provided clinically or mechanistically relevant information on obesity pharmacotherapy, T2DM, obesity-related metabolic complications, treatment-response heterogeneity, multi-omics, or precision-treatment strategies. Publications were excluded when they were outside the scope of the review, duplicated more complete or more recent evidence, or provided insufficient information on the population, intervention, or outcomes. Preclinical evidence was included selectively when required to explain the mechanism of an emerging pharmacological platform.

The pharmacological scope included established therapies used as contextual comparators, recently approved treatments, agents under regulatory review, and investigational compounds in phase 1, 2, or 3 development for obesity, T2DM, or obesity-related metabolic complications. Because this was a narrative review rather than a systematic review or meta-analysis, no formal protocol registration, duplicate study selection, risk-of-bias assessment, or pooled quantitative synthesis was performed. Clinical trial registry records, regulatory documents, and company communications were used primarily to describe current development or regulatory status and preliminary findings not yet available in peer-reviewed form. Such sources are explicitly distinguished from peer-reviewed randomized clinical evidence and are interpreted as provisional until full data are published. This approach improves transparency but does not eliminate the possibility of selection bias inherent to narrative reviews.

As it was indicated above, T2DM is not a single disease entity but a heterogeneous syndrome that may arise through different combinations of IR, beta-cell dysfunction, adipose tissue dysfunction, hepatic steatosis, inflammation, altered incretin signaling, gut microbiome changes, and neurobehavioral drivers of food intake. Data-driven clustering studies have identified subgroups of adult-onset DM with differing risks of complications and potentially different therapeutic needs [16]. Genetic and multi-trait analyses further support the presence of mechanistically distinct pathways leading to T2DM, including insulin secretion, insulin action, adiposity distribution, lipid metabolism, and hepatic metabolic traits [22].

In clinical practice, obesity-associated T2DM commonly coexists with visceral adiposity, hepatic IR, MASLD, dyslipidemia, arterial hypertension, OSA, and increased CV risk. Two patients with the same BMI and HbA1c may nevertheless differ substantially in beta-cell reserve, liver fat, inflammatory activity, appetite biology, skeletal muscle status, kidney function, CV vulnerability, and treatment barriers. These differences are clinically relevant, but most have not been validated as predictive biomarkers for selecting one anti-obesity pharmacotherapy over another [16, 17, 22].

Accordingly, current treatment selection should continue to integrate established clinical factors, including glycemic status, obesity severity, cardiorenal and hepatic comorbidities, risk of hypoglycemia, gastrointestinal (GI) tolerability, frailty, nutritional status, patient preference, and access. Molecular or multi-omics profiles may eventually refine this process, but phenotype-treatment associations discussed in this review should be interpreted as hypothesis-generating unless they have been prospectively validated for differential treatment response.

Recent recommendations emphasize management of the cardiometabolic continuum beyond glucose-centered care, including body weight, comorbidities, and long-term patient-centered outcomes [3]. In parallel, the pharmacological landscape of T2DM and obesity has expanded rapidly. When interpreting this evidence, metabolic efficacy endpoints such as weight reduction and HbA1c should be distinguished from intermediate cardiometabolic risk markers and from CV or kidney event reduction demonstrated in dedicated outcome trials.

Multi-omics refers to the integrated analysis of multiple biological layers, including the genome, epigenome, transcriptome, proteome, metabolome, lipidome, and microbiome [23]. In T2DM, these technologies may help move from broad disease labels toward molecularly informed sub-phenotypes and treatment-response signatures [24–26]. Their relevance is particularly strong in obesity-associated T2DM, where metabolic dysfunction involves several interacting organs and pathways rather than isolated hyperglycemia [6, 10, 13, 14, 27].

Genomics can identify inherited variation associated with insulin secretion, adiposity distribution, hepatic lipid metabolism, drug metabolism, and susceptibility to adverse events. Although most common genetic variants individually have modest predictive value, pathway-level genetic information may improve subtype classification and reveal mechanisms that influence drug response [28]. For example, genetic evidence has identified clusters of T2DM risk loci linked to beta-cell dysfunction, proinsulin processing, obesity, lipodystrophy-like fat distribution, and liver-lipid traits [22].

Epigenomics may capture the influence of obesity, hyperglycemia, aging, inflammation, and environmental exposures on metabolic tissues [25, 29]. DNA methylation and chromatin accessibility patterns could eventually help explain metabolic memory, differential reversibility of IR, or long-term response to therapy for weight reduction [30]. Transcriptomics and proteomics may provide functional information on inflammatory pathways, adipose tissue remodeling, hepatic stress, beta-cell stress, endothelial dysfunction, and muscle metabolism [25, 29].

Metabolomics and lipidomics can provide direct information on small-molecule and lipid signatures that reflect the current metabolic state [31, 32]. Branched-chain amino acids, acylcarnitines, bile acids, triglyceride-rich lipoprotein patterns, ceramides, sphingolipids, ketone-related metabolites, and markers of hepatic lipid flux have been associated with IR, lipotoxicity, and MASLD-related phenotypes [33]. Treatment-related changes in these profiles may provide mechanistic or pharmacodynamic information; however, they should not be interpreted as validated pretreatment predictors of response or safety unless prospectively confirmed.

The gut microbiome represents another relevant omics layer because incretin secretion, bile acid metabolism, intestinal barrier function, immune activation, and energy harvest may be influenced by microbial ecology [34]. Although microbiome-guided prescribing is not yet ready for routine T2DM care, microbiome signatures may help explain interindividual variability in GI tolerability, appetite response, and metabolic improvement during incretin-based therapy [35, 36].

The ultimate value of multi-omics will depend on integration and validation rather than on the number of measured molecular layers. Artificial intelligence (AI) and multimodal computational models may support integration of clinical, imaging, digital, and multi-omics data, but their clinical utility depends on interpretability, reproducibility, calibration, external validation, standardized assays, and equitable implementation [37–39]. Throughout this review, three levels of evidence are distinguished: established clinical evidence from randomized trials, regulatory studies, or routine practice; mechanistic or biomarker evidence that may explain heterogeneity but does not constitute a prescribing criterion; and proposed future multi-omics-guided treatment selection, which remains investigational. At present, no validated multi-omics algorithm can reliably select among individual anti-obesity pharmacotherapies in patients with T2DM and obesity.

Energy balance and glucose homeostasis are regulated through a network linking the gut, central nervous system (CNS), liver, adipose tissue, skeletal muscle, pancreas, kidney, and CV system. Gut-derived hormones such as GLP-1, GIP, glucagon, peptide YY, and amylin modulate appetite, gastric emptying, insulin secretion, glucagon secretion, hepatic metabolism, and satiety signaling [40–46]. GLP-1 receptor activation reduces appetite, slows gastric emptying, improves glucose-dependent insulin secretion, and lowers glucagon secretion. GIP receptor signaling influences insulin secretion and adipose tissue biology, and its pharmacological role appears context-dependent. Glucagon receptor agonism may increase hepatic energy expenditure and lipid oxidation, which is relevant to MASLD/MASH phenotypes but may also require careful monitoring of heart rate (HR), tolerability, and glycemic effects [47–49]. Amylin contributes to meal termination, satiety, and potentially reward-related eating behavior [45, 46].

From a systems-biology perspective, emerging therapies modulate inter-organ pathways rather than acting solely through appetite suppression. Multi-omics profiling may help map gut-brain-liver-adipose tissue-skeletal muscle communication by identifying tissue-specific and circulating signatures of response. For example, high liver fat, inflammatory activity, or a lipotoxic metabolomic profile provides a mechanistic rationale for investigating whether therapies with substantial weight and hepatic effects produce differential benefit in particular subgroups. Similarly, sarcopenic obesity provides a rationale for prioritizing preservation of skeletal muscle health and resistance exercise. These examples are hypothesis-generating and have not been validated as treatment-selection criteria; prospective biomarker-enriched trials are required to test treatment-by-biomarker interactions.

For clarity, biomarkers in this review are classified according to clinical maturity. Established clinical biomarkers or variables are routinely used in care, although none currently constitutes an omics-based tool for selecting a specific anti-obesity pharmacotherapy. Candidate biomarkers are molecular or metabolic features supported by human treatment-response data but not prospectively validated for treatment selection. Exploratory associations are signals identified mainly in cross-sectional, post hoc, or mechanistic analyses that require independent replication. Hypothetical future biomarkers are biologically plausible markers proposed from systems biology or pathway-level reasoning but currently lack sufficient clinical evidence. At present, no genomic, metabolomic, lipidomic, proteomic, microbiome, or integrated multi-omics signature is validated for selecting among GLP-1 receptor agonists, dual or triple receptor agonists, amylin-based therapies, or other emerging anti-obesity medications (AOM) in patients with T2DM and obesity.

Established clinical variables that inform treatment decisions include BMI, waist circumference (WC), duration of T2DM, HbA1c, beta-cell reserve, kidney function, CVD, MASLD status, eating behavior, medication history, and previous intolerance. Genomic, epigenomic, transcriptomic, proteomic, metabolomic, lipidomic, and microbiome data may add biological resolution, but the available evidence is heterogeneous. Metabolomic and lipidomic signatures related to IR, lipotoxicity, hepatic dysfunction, or branched-chain amino acid metabolism are best regarded as candidate or exploratory markers unless prospective treatment-by-biomarker interactions have been demonstrated [22, 24–33, 102]. Microbiome signatures associated with bile-acid metabolism, incretin secretion, GI tolerability, or metabolic response remain exploratory or hypothetical and are not validated prescribing tools [34–36].

Pharmacometabolomics illustrates the distinction between association and clinical utility. Cross-sectional studies by Naja and colleagues identified metabolite patterns associated with good or poor response to metformin and sulfonylureas [24, 26], but these signatures have not been prospectively validated for choosing therapy. Pharmacogenetic evidence is similarly promising but incomplete. For example, SLC2A2 rs8192675 has been associated with differential HbA1c response to metformin [21], and GLP1R rs10305420 has been associated with glycemic response to liraglutide in one clinical population [101]. These findings are candidate biomarkers rather than established prescribing criteria.

Predictive biomarkers should also be distinguished from pharmacodynamic biomarkers. A predictive biomarker should identify, before treatment initiation, differential benefit from one therapy compared with another. By contrast, a pharmacodynamic marker reflects a biological change after treatment exposure and may clarify mechanism without supporting pretreatment drug selection. Exploratory metabolomic and lipidomic changes during tirzepatide therapy, including changes in branched-chain amino acid-related metabolites and lipid species, are examples of pharmacodynamic signatures rather than validated pretreatment predictors [103]. Similarly, preserved C-peptide secretion may predict some clinical responses to GLP-1 receptor agonists in insulin-treated T2DM, but the available evidence remains limited and does not constitute an omics-based selector [114]. Table 3 summarizes representative examples and their validation status.

Combination and sequencing strategies differ substantially in evidentiary maturity. Some combinations are supported by clinical or real-world evidence, whereas phenotype-directed or omics-directed combinations remain hypothesis-generating. This distinction is important because a combination may be clinically reasonable on the basis of established cardiorenal or glycemic indications even when no biomarker-guided selection strategy has been validated.

Metformin remains foundational therapy in many treatment pathways. Sodium-glucose cotransporter 2 (SGLT2) inhibitors have demonstrated CV and kidney benefits in appropriate T2DM populations and may be selected according to established CKD, HF, or CV indications [115–122]. These dedicated outcome data should be distinguished from surrogate improvements in body weight, HbA1c, BP, lipids, or albuminuria. Thiazolidinediones may remain useful in selected insulin-resistant or MASLD-predominant patients because of effects on insulin sensitivity and hepatic steatosis, although weight gain, edema, and HF risk limit broader use [123, 124]. Incretin therapy combined with basal insulin and cagrilintide combined with semaglutide are additional examples supported by randomized clinical data [89, 125].

By contrast, preferential use of pioglitazone according to a multi-omics IR signature, glucagon-containing therapies according to a hepatic lipidomic profile, or amylin-based therapy according to a hyperphagic molecular phenotype remains investigational. These strategies are mechanistically plausible but require prospective comparative studies showing treatment-by-biomarker interactions.

Similarly, induction-maintenance strategies, treatment de-escalation, and sequencing after substantial weight reduction require dedicated trials. Future precision algorithms should only be implemented if they improve patient-relevant outcomes beyond high-quality conventional selection based on glycemic status, CV and kidney indications, hepatic disease, adverse events risk, functional status, patient preference, and access.

The pharmacological landscape of T2DM with obesity has changed rapidly, with therapies producing effects across several distinct domains. These domains should not be conflated: body weight reduction and glycemic improvement are metabolic efficacy endpoints; BP, lipid profile, liver fat, inflammation, and albuminuria are intermediate or surrogate measures; and CV death, MACE, HF events, CKD progression, and kidney failure are clinical outcomes requiring dedicated outcome evidence.

The current evidence supports a clear distinction between population-level therapeutic efficacy and precision treatment selection. Randomized clinical trials have established that several incretin-based and emerging anti-obesity therapies can produce substantial average improvements in body weight and glycemic control, and selected established drug classes have dedicated CV or kidney outcome evidence. However, these findings do not establish that genomic, metabolomic, lipidomic, proteomic, or microbiome profiles can currently select one anti-obesity therapy over another in an individual patient.

This distinction is central to the interpretation of the proposed framework. Biological features such as hepatic steatosis, IR, appetite dysregulation, inflammation, or sarcopenic obesity may provide a mechanistic rationale for studying differential response to glucagon-containing, amylin-based, incretin-based, or body-composition-oriented strategies. Mechanistic plausibility, treatment-induced molecular changes, and retrospective biomarker associations are nevertheless insufficient to establish predictive utility. A further distinction is required between metabolic efficacy and organ-specific clinical benefit: improvements in body weight, HbA1c, BP, lipid profile, liver fat, or albuminuria may indicate favorable risk modification but cannot be assumed to reduce CV or kidney events without dedicated outcome evidence.

Predictive biomarkers must also be separated from pharmacodynamic signatures. Many omics studies identify molecular changes during treatment or correlate baseline features with response, but such findings do not necessarily predict, before treatment, which therapy will provide greater benefit than an alternative. Clinical precision prescribing requires prespecified treatment-by-biomarker interactions, replication in independent and diverse cohorts, standardized and reproducible assays, calibration and external validation of prediction models, and demonstration of incremental value beyond conventional clinical variables.

The principal contribution of this review is therefore not the introduction of a new omics technology or disease classification, but the integration of contemporary anti-obesity pharmacology with an explicit translational hierarchy of evidence. By separating established population-level effects from candidate molecular predictors, exploratory pharmacodynamic signatures, and hypothetical phenotype-treatment associations, the framework clarifies where precision pharmacotherapy currently stands and what evidence is still required before multi-omics can influence routine prescribing in T2DM with obesity.

Clinical translation will require more than statistically significant associations. Multi-omics models must resist overfitting, remain calibrated and transportable across populations and platforms, and be supported by standardized assays and transparent reporting. Most importantly, prospective studies must show that algorithm-guided decisions improve patient-relevant outcomes, safety, persistence, or efficiency compared with high-quality conventional care. Until these requirements are met, multi-omics should be considered a research framework for generating and testing precision-treatment hypotheses rather than an established component of routine prescribing. The major challenge is no longer simply to identify additional molecular associations, but to determine which signals are predictive, reproducible, clinically actionable, and sufficiently informative to change treatment decisions beyond established clinical assessment.

First, future clinical trials should move beyond average treatment effects and incorporate prespecified clinical, metabolic, imaging, and multi-omics phenotyping. Candidate biomarkers should be defined before analysis and tested for treatment-by-biomarker interactions rather than merely correlated with outcome. Replication in independent cohorts is necessary to determine whether a marker is truly predictive of differential treatment response rather than simply prognostic or mechanistically plausible.

Second, trials should deliberately include clinically relevant T2DM subgroups, including early and long-duration disease, MASLD/MASH, CKD, HF, older adults, sarcopenic obesity, and participants from diverse ancestry and socioeconomic groups. External validity requires that predictive signatures be reproducible across populations, laboratories, platforms, and healthcare settings [17, 47, 130–135].

Third, future studies should prespecify distinct outcome domains. Metabolic efficacy should include body weight, HbA1c, time in range, insulin or sulfonylurea de-intensification, and beta-cell function. Intermediate organ or risk markers may include liver fat, MASH resolution, BP, lipids, or albuminuria. CV and kidney protection should be reserved for adjudicated clinical outcome evidence from appropriately designed trials. Body-composition endpoints should distinguish total lean mass from skeletal muscle quantity and should be complemented by muscle strength and physical performance, including validated measures such as handgrip strength, chair-stand performance, gait speed, or the Short Physical Performance Battery [100, 109, 110, 113].

Fourth, AI and computational modeling may facilitate integration of high-dimensional clinical, imaging, digital, and multi-omics data, but methodological rigor is essential before implementation. Because the number of molecular features may greatly exceed the number of participants, multi-omics models are particularly vulnerable to overfitting. Development should therefore include appropriate internal validation followed by independent external validation in temporally, geographically, and demographically distinct cohorts. Discrimination should be accompanied by calibration assessment, because accurate separation of responders from non-responders does not necessarily imply accurate estimation of individual treatment benefit. Reproducibility may also be compromised by batch effects, sample handling, storage conditions, platform differences, reagent lots, preprocessing, normalization, and analytical pipelines. Harmonized pre-analytical procedures, assay standardization, quality-control protocols, and transparent reporting are prerequisites for transportability [23, 38, 39, 104, 136, 137].

Fifth, statistical validity alone is insufficient to establish clinical utility. Candidate algorithms should be tested prospectively to determine whether biomarker- or multi-omics-guided treatment selection improves efficacy, safety, treatment persistence, patient-reported outcomes, or resource use compared with high-quality conventional assessment. Combination and sequencing strategies likewise require dedicated randomized trials; clinically supported combinations should be distinguished from phenotype-directed hypotheses that remain unvalidated.

Finally, pediatric and adolescent populations remain an important evidence gap. Trials should evaluate not only weight and glycemic outcomes but also growth, puberty, bone health, nutritional adequacy, psychological outcomes, body composition, muscle function, and durability after treatment discontinuation. Precision approaches in younger populations should not be extrapolated directly from adult omics or pharmacotherapy data.

Coexisting obesity and T2DM represent a heterogeneous metabolic phenotype requiring treatment strategies that extend beyond glycemic control and body weight reduction alone (Figure 1). Emerging incretin-based, amylin-based, multi-receptor, and oral therapies, together with approaches targeting body composition, have broadened the therapeutic landscape. Their selection, however, remains guided primarily by conventional clinical characteristics, approved indications, comorbidities, safety, and patient preferences rather than by validated molecular profiles.

Multi-omics may ultimately refine patient stratification and treatment selection, but its current role remains investigational. Translation into routine precision pharmacotherapy will require prospectively validated predictive biomarkers, standardized assays, externally validated and calibrated algorithms, direct comparative evidence, and demonstration that biomarker-guided decisions improve clinically relevant outcomes beyond standard care.