GLP-1: nutrient sensing, DPP-4 and the gut–brain system

Glucagon-like peptide-1 links nutrient arrival in the intestine with insulin secretion, gastric motility and food intake. Its biology explains why glucose taken by mouth can elicit more insulin than intravenously delivered glucose at a matched glucose concentration. It also explains the interest in foods that stimulate enteroendocrine L cells. These physiological signals must be distinguished from the sustained exposure, doses and clinical outcomes of receptor agonist medicines. A rise in endogenous GLP-1 after a meal does not establish equivalent weight loss or disease protection.

One precursor, different tissues

The human GCG gene, at 2q24.2, encodes preproglucagon. The canonical precursor includes a signal peptide; the commonly described 160-residue proglucagon is the chain after removal of that signal, rather than the complete translation product. The gene has a modular exon organization containing glucagon-related peptide sequences. Precise exon boundaries, transcript lengths and cleavage coordinates require a named transcript and numbering system; they should not be treated as a universal six-box genomic map.

Prohormone convertase PC2, encoded by PCSK2, predominates in the conventional pancreatic α-cell processing pathway. It releases glucagon and glucagon-related pancreatic polypeptide, while the C-terminal major proglucagon fragment retains the GLP sequences. In intestinal L cells and brainstem preproglucagon neurons, PC1/3, encoded by PCSK1, supports production of GLP-1, GLP-2 and other proglucagon-derived peptides. Glicentin and oxyntomodulin retain the glucagon sequence within larger peptides; oxyntomodulin has biological activity rather than being simply “blocked glucagon.” α-cell processing can change during injury or metabolic stress, so the division is a useful dominant pattern, not an immutable tissue prohibition.

The active native forms are GLP-1(7–36)amide and GLP-1(7–37), containing 30 and 31 residues respectively. Their numbering refers to the GLP-1 segment, not residue seven of the whole precursor. The N-terminal histidine is important for receptor activation. GLP-1(1–37) has much weaker conventional incretin activity; describing its processing as a single unidentified obligatory enzyme step conceals the specificity of prohormone maturation.

Peptidylglycine α-amidating monooxygenase, PAM, converts a glycine-extended peptide into an amidated peptide. Its hydroxylating and lyase activities involve oxygen, copper and reducing equivalents supplied by ascorbate; glyoxylate is released. The following is a processing scheme, not a fully balanced chemical equation:

$$\mathrm{GLP\!\!-1}(7\!\!-37)\xrightarrow{\mathrm{PAM}}\mathrm{GLP\!\!-1}(7\!\!-36)\mathrm{amide}+\mathrm{glyoxylate}.$$

The amidated form predominates in many human measurements, often around four fifths of the active pool. The proportion depends on the assay and sampling. Both active forms can activate GLP-1R; predominance alone does not establish a clinically superior peptide or an exact universal distribution. Tissue processing and early incretin physiology are described by Drucker (2006), Holst (2007) and Baggio and Drucker (2007).

GLP-1R: ligand recognition and signaling

GLP-1R is a class B1 G-protein-coupled receptor with an extracellular N-terminal domain, seven transmembrane helices, extracellular and intracellular loops, and a cytoplasmic C-terminal region. In the two-domain model, the peptide C-terminal region interacts with the extracellular domain, while its N-terminal region engages the transmembrane core. These contacts stabilize an active conformation in which the intracellular portion of TM6 moves outward to accommodate Gs. They are structural interactions, not a catalytic active site in the peptide.

Structural identifiers need careful interpretation. Zhang and colleagues’ 2017 PDB 5VAI is a GLP-1–Gs complex with a rabbit receptor and components from several species. PDB 6B3J is a biased peptide agonist-bound human GLP-1R–Gs complex; PDB 6ORV contains the nonpeptide agonist TT-OAD2. They are not three equivalent native human GLP-1 snapshots. Exact residue contacts, displacement distances and “ionic locks” must be linked to the relevant construct and comparison. Liang and colleagues’ generic 2017 class B complex paper concerns another class B receptor and cannot substitute for a GLP-1R structure.

Ligand activation facilitates GDP–GTP exchange on Gαs and adenylyl cyclase activation. ATP supplies cyclic AMP:

$$\mathrm{ATP}\xrightarrow{\mathrm{adenylyl\ cyclase}}\mathrm{cAMP}+\mathrm{PP_i}.$$

cAMP activates PKA and the exchange factor Epac2, encoded by RAPGEF4. PKA modulates ion channels, exocytotic proteins and transcriptional responses, including CREB phosphorylation. Epac2 activates Rap1 and participates in granule recruitment and priming with proteins such as RIM2, Piccolo, Rab3 and Munc13. PLCε and intracellular calcium release contribute in particular experimental settings. This machinery increases secretory competence rather than making calcium dispensable.

GRKs and β-arrestins regulate desensitization, internalization and trafficking. Arrestins can organize additional signaling, including ERK-related responses, but receptor internalization does not necessarily terminate every Gs signal: endosomal signaling and ligand bias matter. A universal linear “arrestin binds, all cAMP stops, independent Akt begins” rule overstates the evidence.

The incretin effect and pancreatic integration

McIntyre and colleagues and Elrick and colleagues independently demonstrated the oral–intravenous insulin difference in 1964. A contemporary matched comparison uses intravenous glucose adjusted to reproduce the oral glucose profile, rather than assuming that one fixed infusion rate always produces an isoglycemic comparison. Using comparable insulin or C-peptide areas under the curve:

$$\mathrm{Incretin\ effect}(\%)=\frac{\mathrm{AUC}_{\mathrm{oral}}-\mathrm{AUC}_{\mathrm{IV}}}{\mathrm{AUC}_{\mathrm{oral}}}\times100.$$

AUC denotes the area over the specified time window; the same baseline convention and analyte must be used in both tests. The oral AUC is the denominator, so a twofold oral response corresponds to 50%, not a 200% incretin effect. Values around 50–70% occur in suitable healthy-study conditions but vary with the glucose load and protocol. GLP-1 and glucose-dependent insulinotropic polypeptide, GIP, both contribute. GIP-producing K cells are concentrated proximally; GLP-1-producing cells occur throughout the intestine with a distal enrichment. Type 2 diabetes often reduces the incretin effect through impaired β-cell responsiveness, particularly to GIP, and other disease-related changes. It is not uniformly explained by a deficient GLP-1 pool or one threshold below 20%.

Glucose supplies the trigger; GLP-1 amplifies it

In β cells, glucose uptake and glucokinase-dependent metabolism raise the ATP/ADP ratio. Human glucose transport differs from the common rodent GLUT2 model: GLUT1/GLUT3 are important in human β cells. Glucokinase is EC 2.7.1.2, rather than the source’s generic hexokinase EC 2.7.1.1. ATP-sensitive potassium channels formed by Kir6.2 and SUR1 close, potassium conductance falls, the membrane depolarizes and voltage-dependent calcium channels open. Their contributions and membrane potentials differ between species and preparations.

$$\frac{[\mathrm{ATP}]}{[\mathrm{ADP}]}\uparrow\;\Rightarrow\;K_{\mathrm{ATP}}\mathrm{\ closure}\;\Rightarrow\;\mathrm{depolarization}\;\Rightarrow\;\mathrm{Ca}^{2+}\mathrm{\ entry}.$$

GLP-1R–cAMP signaling amplifies this glucose-supported process. PKA and Epac2 modify calcium sensitivity and granule availability. Readily releasable and reserve granule pools help explain early and sustained secretory phases. SNARE proteins including syntaxin, SNAP-25 and VAMP2, calcium sensors such as synaptotagmin, and priming machinery participate, without every proposed protein contact being obligatory in every cell.

Glucose dependence supports a low intrinsic hypoglycemia risk, not an absolute guarantee. The source’s claim of no secretion at euglycemia and zero risk under every condition is too strong. Concomitant insulin or sulfonylureas can alter clinical risk; fixed thresholds of 4.5–5.0 or 3.8–4.0 mmol/L do not describe all patients and experimental preparations.

GLP-1 also suppresses glucagon in appropriate glucose conditions. Somatostatin released by δ cells acts on α-cell SSTR2 and is an important paracrine mechanism. Insulin, local signaling and possible direct effects must be considered according to the model. Gi-coupled inhibition of secretion is not proof of a complete instantaneous blockade of all α-cell calcium channels. Preserving counterregulation at low glucose is a physiological tendency rather than a universal numerical safety switch.

PKA/CREB, Akt, Bcl-2-family proteins and PDX1 link receptor signaling with insulin production, cell survival and differentiated β-cell function in experimental systems. CREB activation generally involves phosphorylation and cofactor control; the source’s “dephosphorylated active CREB” statement reverses its own preceding pathway. Rodent proliferation and possible neogenesis findings do not establish regeneration of adult human β-cell mass by food, supplements or medicines.

DPP-4, degradation and assay interpretation

DPP-4/CD26, EC 3.4.14.5, is a type II membrane serine peptidase that also occurs in soluble form. Its glycosylated monomers form a dimer; the extracellular region contains an eight-bladed β-propeller and α/β-hydrolase domain. Rasmussen and colleagues’ structure 1N1M contains an inhibitor/substrate analogue. Its correct associated PubMed identifier is 12483204, not the source’s 12483274. The human catalytic triad is Ser630–Asp708–His740. Substrate recognition strongly favors N-terminal peptides with penultimate proline or alanine, but “exclusively” is an unnecessarily absolute specificity claim.

Mentlein and colleagues (1993) and Deacon and colleagues (1995) established rapid N-terminal degradation. DPP-4 hydrolyzes the bond after Ala8, releasing His7–Ala8:

$$\mathrm{GLP\!\!-1}(7\!\!-36)\mathrm{amide}+\mathrm{H_2O}\xrightarrow{\mathrm{DPP\!\!-4}}\mathrm{His\!\!-Ala}+\mathrm{GLP\!\!-1}(9\!\!-36)\mathrm{amide}.$$

Serine attack first forms a tetrahedral intermediate, which collapses to an acyl-enzyme; subsequent water-mediated hydrolysis releases the product. A tetrahedral intermediate and an acyl-enzyme are distinct stages, rather than one “tetrahedral acyl-enzyme.” The truncated peptide has greatly reduced conventional incretin activity. Antagonism and receptor-independent cardiovascular/metabolic actions have been studied at particular concentrations and models; neither a universal 1,000-fold affinity change nor physiologically important subnanomolar antagonism follows from the source.

Intact native GLP-1 has a circulating half-life of roughly one to two minutes. Rapid degradation in intestinal capillaries, liver and circulation limits intact systemic exposure. Frequently cited estimates that only about 10–15% of secreted intact peptide reaches systemic blood are model-dependent estimates, not a fixed fraction for every human meal. Sequential extraction must specify whether each percentage refers to the starting pool or to the remaining amount; multiplying two remaining fractions differs from subtracting both from the initial pool.

For an idealized first-order decay after secretion stops:

$$C(t)=C_0e^{-kt},\qquad t_{1/2}=\frac{\ln2}{k}.$$

C is intact peptide concentration, t is time and k has inverse-time units. Ongoing secretion and distribution prevent this simple equation from describing every postprandial sample. An isolated in-vitro Michaelis constant does not determine in-vivo clearance. The claimed requirement for hundreds of pmol/kg/min native infusion is unsupported; human experimental infusions have operated on a much lower scale. No infusion protocol is prescribed here.

Neprilysin/NEP/CD10, EC 3.4.24.11, can cleave internal peptide bonds. Kidneys contribute to elimination of peptide and metabolites; chronic kidney disease changes exposure. Exact cleavage sites, compulsory megalin–cubilin uptake and complete renal elimination of every fragment require separate evidence rather than a universal diagram. “Total GLP-1” and “active GLP-1” assays measure different pools. Collection conditions and inhibition of ex-vivo degradation matter; one concentration cannot directly quantify L-cell number, receptor responsiveness or satiety.

L-cell nutrient sensing

Open-type L cells contact the lumen through an apical process and release hormone from the basolateral side. They are enriched distally but also present in proximal intestine; some coexpress PYY or other hormones. Receptors are not all restricted to apical microvilli. Nutrient absorption, basolateral access, species, segment and preparation alter the apparent sensor distribution.

Microbial short-chain fatty acids

Microbial fermentation of otherwise poorly digested carbohydrate generates acetate, propionate and butyrate. Luminal concentrations, proportions and circulating exposure differ substantially. Tolhurst and colleagues (2012) showed FFAR2-dependent GLP-1 secretion in experimental systems. FFAR2/GPR43 couples to Gq/11 and Gi/o; Gq–PLC–IP3 signaling can release intracellular calcium. FFAR3/GPR41 predominantly couples to Gi/o. Its activity should not be equated to a universal stimulatory cAMP pathway. Lin and colleagues (2012) demonstrated relevant effects independent of FFAR3, supporting multiple mechanisms rather than one required receptor.

Butyrate can inhibit histone deacetylases and influence epithelial metabolism and differentiation. NEUROG3 and NEUROD1 belong to enteroendocrine developmental programs. These findings make altered L-cell capacity plausible in experimental contexts, but do not establish that every fiber dose specifically increases H3K9ac/H3K27ac at the human GCG promoter or causes clinically useful L-cell hyperplasia. Acute hormone release and long-term changes in cell differentiation are different endpoints.

Fatty acids, monoacylglycerols and OEA

FFAR1/GPR40 and FFAR4/GPR120 sense long-chain fatty acids in relevant preparations. Hirasawa and colleagues (2005) helped establish FFAR4-related secretion. Gq/calcium signaling and β-arrestin-linked anti-inflammatory pathways are experimentally important, but FFAR4–TAB1–TAK1 observations in other cell types cannot automatically be assigned to every human L cell.

GPR119 senses lipid-derived ligands including 2-monoacylglycerols and oleoylethanolamide, OEA; Lauffer and colleagues (2009) showed OEA-dependent secretion in L-cell models. Gs–cAMP signaling is distinct from Gq–IP3 signaling. Increasing cAMP amplifies exocytosis but does not demonstrate that secretion occurs independently of calcium. Dietary oleic acid, enteroendocrine ligand production and a bottle of olive oil are not interchangeable pharmacological agonist definitions. OEA is synthesized through lipid intermediates and also has PPARα-associated satiety biology; its formation is not a simple one-step oleic acid conversion confined to GPR119.

Sugars, peptides, amino acids and taste receptors

SGLT1/SLC5A1 provides electrogenic sodium-coupled glucose and galactose transport. The conventional transport stoichiometry is two sodium ions per glucose molecule:

$$2\mathrm{Na}^{+}_{\mathrm{lumen}}+\mathrm{glucose}_{\mathrm{lumen}}\xrightarrow{\mathrm{SGLT1}}2\mathrm{Na}^{+}_{\mathrm{cell}}+\mathrm{glucose}_{\mathrm{cell}}.$$

The sodium gradient, maintained indirectly by Na/K-ATPase, supports inward transport and depolarization. Calcium entry can trigger secretion. Gorboulev and colleagues (2012) established its importance in glucose absorption and incretin responses. GLUT2-mediated uptake and sweet-taste TAS1R2/TAS1R3 pathways vary with the model. Inhibiting proximal transport may blunt an early response while changing distal delivery and later secretion, so it need not abolish all GLP-1 release.

PepT1/SLC15A1 transports di- and tripeptides with protons and can contribute to electrogenic sensing. CaSR, a class C GPCR, responds to extracellular calcium and amino-acid modulation, especially aromatic amino acids in appropriate conditions. Glutamine, arginine and protein hydrolysates can affect secretion through transport, metabolism and receptor mechanisms; assigning all of them exclusively to one CaSR binding pocket overstates certainty.

Gut taste-signaling studies, including Kokrashvili and colleagues (2009), implicate gustducin, PLCβ2, IP3 and TRPM5. TRPM5 conducts monovalent cations and can promote depolarization following a calcium signal; it is not itself a calcium entry channel. Bitter TAS2R and sweet TAS1R effects depend on receptor, compound and preparation. A bitter food or polyphenol is not thereby a demonstrated effective human GLP-1 treatment.

Gut–brain circuits and gastric motility

Vagal sensory neurons, the nodose ganglion, the nucleus of the solitary tract, NTS, and other brainstem circuits integrate intestinal, mechanical and metabolic information. GLP-1R-expressing vagal populations have distinct functions, including mechanosensory roles. All such neurons should not be drawn as identical terminals directly downstream of an L cell.

Brainstem preproglucagon neurons synthesize GLP-1 locally and project to hypothalamic and other regions. Trapp and Cork (2015) describe this central system. Brierley and colleagues (2021) experimentally demonstrated that central and peripheral GLP-1 systems can suppress eating through independent circuits. This directly limits the source’s obligatory L-cell→vagus→PPG→hypothalamus chain.

The area postrema and median eminence provide specialized access regions. Drug-dependent brain access, exposure duration, receptor distribution and tissue barriers matter; fenestrated capillaries do not imply unrestricted native-peptide penetration throughout the brain. OVLT should not be promoted to a universal established portal for GLP-1 anorexia without specific evidence. Sisley and colleagues (2014) and Secher and colleagues (2014) identify neuronal and arcuate mechanisms for liraglutide in experimental settings, not identical pathways for every endogenous peptide pulse.

POMC/CART, α-MSH–MC4R and NPY/AgRP networks integrate energy signals. Their anatomy includes direct and indirect actions, inhibitory interneurons and projections beyond the arcuate nucleus. This network also interacts with leptin signaling and resistance; common circuit components do not make leptin and GLP-1 equivalent hormones. Alhadeff and colleagues (2012) demonstrated central GLP-1 projections to VTA and nucleus accumbens involved in food intake. Reward modulation is not complete elimination of dopamine, hedonic appetite or “food noise.” That informal phrase does not establish a specific synapse or universal treatment response.

GLP-1 can slow gastric emptying through vagal and enteric integration, gastric accommodation and pyloric coordination. Gastric distension adds mechanosensory information. Intraganglionic laminar endings, IGLEs, and intramuscular arrays, IMAs, are recognized vagal mechanoreceptor terms; “Intraneural Galvanic Mechanoreceptors” is not an appropriate anatomical category. The effect varies with baseline emptying, agent and exposure, and can diminish with sustained stimulation. Slower emptying, reduced energy intake and nausea are related but distinct observations.

Beyond pancreas and appetite

Experimental PKA/AMPK/eNOS/NO pathways offer vascular hypotheses, and mPTP-related models investigate ischemic injury. Receptor detection in every endothelial cell, vascular smooth-muscle cell or cardiomyocyte cannot be assumed from nonvalidated antibodies. Clinical cardiovascular benefits of selected medicines must be established in outcome trials, not inferred from one phosphorylation site or transferred to nutrients.

Crajoinas and colleagues studied PKA-associated NHE3/SLC9A3 regulation and renal sodium handling. Tubular transport findings do not settle precise human GLP-1R localization: vascular and indirect mechanisms remain relevant. Natriuresis, blood pressure change and long-term kidney outcomes are different levels of evidence. Species-specific phosphosites and apical receptor localization need separate confirmation.

NF-κB and NLRP3-related experimental findings support investigation of inflammatory responses. They do not prove ubiquitous functional GLP-1R in every macrophage. NLRP3–caspase-1 processing directly concerns IL-1β and IL-18; IL-6 and TNF are not all products cleaved by that complex. Changes in weight, glucose and tissue stress can contribute indirectly.

Hepatic effects likewise include altered insulin/glucagon signaling, reduced adipose fatty-acid delivery and changes in energy intake. Hepatocyte GLP-1R expression remains a methodological concern. SREBP-1c and AMPK hypotheses should not be presented as proof that native GLP-1 or supplements cure MASH. Newsome and colleagues’ 2021 phase 2 trial showed more steatohepatitis resolution with semaglutide, but did not establish a significant between-group fibrosis-stage improvement. Histology, liver fat, transaminases and clinical outcomes are separate endpoints.

Medicines and the limits of nutritional comparisons

Liraglutide and semaglutide are GLP-1R agonists. Semaglutide combines a DPP-4-resistant amino-acid substitution and C18 fatty-diacid-associated albumin binding; liraglutide uses a C16 acyl chain. The roughly one-week half-life belongs to semaglutide, not all agonists. Tirzepatide acts at GIP and GLP-1 receptors; retatrutide targets GIP, GLP-1 and glucagon receptors. Receptor identities should not be collapsed into “three brain satiety receptors.”

These are published trial regimens for comparison, not individual dosing instructions. Cross-trial percentages are not a randomized comparison of medicines. Current approvals, formulations and prescribing decisions require the applicable current product information.

DPP-4 inhibitors such as sitagliptin and vildagliptin preserve endogenous active incretins, including GIP. They inhibit enzyme activity rather than necessarily covalently blocking all three catalytic residues. They tend to be weight-neutral and have a low intrinsic hypoglycemia risk, which changes with concomitant treatment. Preserving native hormone does not produce the same exposure or outcome evidence as prolonged receptor agonism; cardiovascular and heart-failure findings also differ among agents.

Nutritional interventions: mechanisms with distinct evidence

Fermentable fiber includes inulin, resistant starch, partially hydrolyzed guar gum and some β-glucans. Fermentability, viscosity, microbiota and tolerance differ. Cani and colleagues’ 2009 GLP-2/permeability mouse work supports a microbiota–metabolic inflammation connection, not guaranteed human L-cell expansion. Increasing fiber can be useful for several nutritional goals, but a universal 15–25 g supplement dose to “maximize GLP-1” is not validated. Bacterial cross-feeding also makes it inaccurate to assign identical butyrate production to Bifidobacterium, Faecalibacterium and Akkermansia.

The ileal brake is a nutrient-triggered feedback system involving motility and hormones, described by Maljaars and colleagues (2008). White-bean Phaseolus vulgaris amylase inhibitors and mulberry Morus alba 1-deoxynojirimycin, DNJ, can alter carbohydrate digestion in relevant products. SGLT1 transports monosaccharides, not undigested oligosaccharides. Delayed digestion may change distal glucose delivery or microbial fermentation; it does not prove every extract drives a powerful SGLT1-mediated GLP-1 wave. Extract standardization, enzyme potency and human trial endpoints matter.

Lu and colleagues’ 2009 berberine experiments in streptozotocin-induced diabetic rats provide a mechanistic lead, rather than demonstration that one universal human AMPK–GCG–microbiota sequence mediates all benefits. Berberine exposure and interactions are discussed separately in the berberine and dihydroberberine review. A generic 500 mg twice-daily “incretin stack” is not derived from a preclinical experiment.

Olive oil contains oleic acid and other constituents; it can be part of a dietary pattern. Lipid-mediated secretion and OEA synthesis do not establish that 15–20 mL is a receptor-targeted therapeutic dose. Additional energy intake must be considered when evaluating weight outcomes.

Protein preloads, including whey, have produced changes in incretin, insulin, gastric emptying and postprandial glucose in small controlled human studies. Protein quantity, timing, baseline diabetes, meal composition and extra energy vary. The source’s 15–20 g at 15–30 minutes is a possible studied strategy family, not a universally validated prescription or complete elimination of the glucose excursion. Amino acids also act directly on insulin and glucagon, so a glucose change cannot automatically be attributed entirely to GLP-1.

Jakubowicz (2014) studied 15 participants in an acute crossover trial with a 50 g whey preload. This small trial does not validate every smaller dose.

Timing, genetics and monitoring

BMAL1, CLOCK and PER-related pathways connect enteroendocrine function with circadian biology in experimental models. Human morning–evening responses also depend on sleep, meal order, composition and glucose tolerance. A universal morning maximum of GCG and SGLT1, or a clock-time prescription for every person, does not follow.

Grant and colleagues (2006) linked TCF7L2 variation, including rs7903146, with type 2 diabetes risk. This does not establish that TT carriers need a higher fiber dose or that a single allele deterministically suppresses intestinal proglucagon. GLP1R variants such as rs10305492 and DPP4 variation may affect particular traits; receptor variation is not automatically altered DPP-4 degradation. Association, function and treatment-response prediction require different evidence.

Monitoring follows an identified clinical question: symptoms, glucose/HbA1c, weight, nutritional adequacy and relevant kidney/liver risks. Postprandial variability and CGM discusses what sensor patterns can and cannot show. CGM does not measure GLP-1 or prove its mediation, and is not obligatory for everyone trying fiber or dietary protein. Routine pancreatic enzyme screening before an unspecified nutritional “stack” is not established. Pancreatitis symptoms, substantial gastrointestinal intolerance and severe delayed gastric emptying warrant clinical evaluation; contraindications depend on the actual medicine or intervention rather than one “forced GLP-1 stimulation” category. Ulcerative colitis is not a blanket contraindication to all listed foods and fibers.

This is a narrative scientific synthesis with targeted verification of disputed structural identifiers, mechanisms and trial results using accessible primary records and publications on 8 October 2026. It is not a systematic review, an independent GRADE assessment, clinical validation or external peer review. Food-induced secretion, experimental pathways and pharmacological outcomes are retained at their respective evidence levels. The useful objective is an appropriate nutritional or clinical outcome, rather than maximizing a hormone measurement by an unvalidated combination.