The Science Behind Test, Don't Guess

Baker and colleagues reviewed how gut bacteria help regulate circulating oestrogen. The main route is an enzyme, β-glucuronidase, that frees oestrogen in the gut so it can be reabsorbed. They propose that lower microbial diversity may reduce this and lower oestrogen, and that this may contribute to several conditions. They also note the effect runs both ways: oestrogen levels shape the microbiome too. This is a narrative review; most of the human evidence it draws on is observational.

Hu and colleagues reviewed how the gut enzyme β-glucuronidase affects oestrogen recycling, and how oestrogen levels in turn shape the gut microbiome. They suggest that measuring this enzyme activity may one day help with early diagnosis of oestrogen-driven cancers. They are clear that it is not yet known which bacteria matter most, and that the evidence is “relatively scattered”. Note the direction: some studies link low gut diversity to lower oestrogen (card above), others to higher. It isn’t settled.

Ervin and colleagues tested 35 gut-bacterial β-glucuronidase enzymes in the lab. Some enzyme classes turned inactive oestrogen back into active oestrone and oestradiol. This was blocked in purified enzymes and mouse stool samples. But blocking one enzyme class did not reduce breast tumours in mice, and the authors conclude that many enzymes are probably involved. It confirms the mechanism exists. It does not show what a stool result means for an individual’s hormone levels.

In mice, removing the ovaries changed the gut microbiome (on a low-fat diet), raised faecal β-glucuronidase activity, and made the gut leakier and more inflamed. Germ-free mice given these animals’ gut bacteria gained more weight and showed liver changes linked to metabolic dysfunction. This suggests the microbiome can contribute to metabolic changes after oestrogen loss. It is a mouse study of sudden surgical oestrogen loss, and it has not been tested in women.

Herman and colleagues reviewed how the brain switches the stress response on and off. Chronic stress can show up as persistently high cortisol, an over-reactive stress response, or a blunted one, depending on the stressor and the person. Chronic stress can also engage different brain circuits from acute stress. Much of this circuitry comes from animal research.

Oyola and Handa reviewed how sex hormones shape the stress (HPA) axis, mostly in rats and mice. In rodents, oestradiol tends to make females’ stress-hormone responses larger, while testosterone tends to dampen males’. The authors note it is unclear how far this applies to humans.

Heck and Handa reviewed sex differences in the stress-hormone response. In adult female rodents, the response to an acute stressor is markedly larger than in males, mainly because of oestradiol and testosterone. The review is mostly about rodents. It includes a brief section on humans and says sex differences after chronic stress are still an open question.

Ring’s review in the American Journal of Medicine describes the many things that can push the stress axis out of balance: psychological stress, diet, disrupted sleep and daily rhythm, environmental exposures, gut health and hormones. It covers integrative treatments including mind–body therapies, diet and lifestyle, and supplements. It notes that cortisol tests “vary in sensitivity and clinical utility”. It is a narrative review, so it describes these approaches rather than grading the trial evidence for each.

In 200 people with ME/CFS and 102 healthy controls, Fluge and colleagues found a pattern of low serum amino acids, mainly in women. The pattern suggested the enzyme that feeds sugar into aerobic energy production (pyruvate dehydrogenase) was being held back. Blood cells showed raised activity of genes that inhibit this enzyme. Muscle cells grown in patients’ serum produced excess lactate. This supports an energy-metabolism problem in ME/CFS. It does not show how to measure it in an individual.

Molnar and colleagues reviewed evidence that mitochondrial dysfunction may contribute to long COVID fatigue, brain fog and exercise intolerance. They draw parallels with ME/CFS and other post-infection syndromes. They list treatments under investigation, such as CoQ10 and NAD+ precursors, and say direct evidence in long COVID is “still emerging”. In their own group of 139 patients, a blood marker of mitochondrial stress was not linked to fatigue severity.

Zong and colleagues reviewed mitochondrial dysfunction across heart disease, neurodegeneration, metabolic syndrome and cancer. They list supplements studied for mitochondrial support, including CoQ10, carnitine, creatine, riboflavin and nicotinamide riboside, alongside drug and experimental treatments. They note that few of the supplement studies are randomised trials; most are open-label.

In a single-centre, single-blind trial in China, 98 adults with migraine and at least one positive food-IgG result were randomised for 12 weeks. One group avoided the foods they tested positive to; the other (sham) group avoided foods they tested negative to. The real-elimination group had fewer migraine days (about one fewer a month than sham), less disability, fewer gut symptoms and better sleep scores. Blood markers of inflammation (IL-6, TNF-α) and the migraine-related peptide CGRP also fell. Mood and anxiety did not differ. The authors call for larger multicentre trials, and the inflammation mechanism they propose is speculative.

In a double-blind trial at 8 US centres, 238 adults with IBS who tested positive to at least one of 18 foods on a new IBS-specific IgG test were randomised to an IgG-guided elimination diet or a sham diet for 8 weeks. 59.6% on the IgG-guided diet reached the main goal (at least 30% less abdominal pain), against 42.1% on the sham diet. The benefit looked larger in constipation-predominant and mixed IBS. The test was a purpose-built 18-food IgG test for IBS made by Biomerica (inFoods IBS), not one of the broad 200-plus-food panels sold for general use, and the authors call for a larger study. Biomerica funded the trial, and four of its authors consult for the company (Gastroenterology 2025, funding and disclosures).

In an open-label study of 73 women with mixed-type IBS, 21 followed a diet guided by IgG food testing, 26 followed a low-FODMAP diet and 26 followed standard advice from their gastroenterologist, all for 8 weeks. Pain-related symptoms improved within the IgG group. Bloating and related symptoms improved within the FODMAP group. The control group showed no significant change. Groups were not randomised, patients knew which diet they were on, and the groups were not compared statistically with each other. So this study cannot show one diet is better than another, despite its title.

In a retrospective review of 407 children attending a Chinese allergy clinic, more than 80% had raised food-IgG4, mostly to egg and milk. Of the 67 who later avoided their IgG4-positive foods for at least three months (about half also took probiotics), IgG4 levels fell. Bigger falls went with more parent-reported improvement. There was no comparison group that did not change their diet, and IgG4 naturally falls when a food is no longer eaten. The study was in children with allergic disease, not adults with food intolerance.

Garmendia and colleagues reviewed what food-specific IgG antibodies mean. IgG to foods is common in people with no symptoms and may be part of normal tolerance. Some IgG-guided elimination diets have improved symptoms in conditions such as IBS. Major allergy organisations advise against routine IgG testing. The authors conclude that IgG testing should currently be treated as investigational and interpreted cautiously alongside the full clinical picture.

Ferritin at or below 30 µg/L identified iron deficiency with 92% sensitivity and 98% specificity in anaemic patients checked against bone marrow or response to iron; the older 12 µg/L cut-off caught only a quarter of them (Mast 1998). In a randomised trial of 144 non-anaemic women with unexplained fatigue, iron reduced fatigue, and the benefit was confined to women with ferritin at or below 50 µg/L (Verdon 2003). Many lab lower limits sit below both figures, which is why a “normal” ferritin can still be the problem. Claims that low-normal ferritin impairs thyroid conversion are mechanistic and untested.

Labs usually flag homocysteine above about 15 µmol/L. Higher levels within that range go with higher cardiovascular and dementia risk in observational studies, which can’t show cause. The clearest trial is VITACOG: in 271 people over 70 with mild memory problems, B vitamins slowed brain shrinkage over two years, and in those starting above 13 µmol/L the rate was 53% lower (Smith 2010). I flag homocysteine at about 11 µmol/L, a clinical judgement rather than a validated cut-off, and read it with B12 and folate to look for a cause.

HOMA-IR estimates insulin resistance from one fasting glucose and insulin. In the original 1985 study it correlated closely with the gold-standard glucose clamp (r = 0.88), but a single estimate was imprecise (coefficient of variation about 31%) (Matthews 1985). Its authors later reviewed its use and warned against reading it as an individual diagnostic test; it is best used for comparing groups and following trends (Wallace 2004). There is no agreed cut-off. In the Whitehall II study, HOMA insulin sensitivity fell steeply over the five years before diabetes was diagnosed, while fasting glucose rose sharply only in the last three (Tabák 2009).

The harmonised definition of metabolic syndrome counts triglycerides of 1.7 mmol/L (150 mg/dL) or more, and HDL below 1.0 mmol/L (40 mg/dL) in men or 1.3 mmol/L (50 mg/dL) in women, as two of its five criteria; three of five make the diagnosis (Alberti 2009). The triglyceride-to-HDL ratio combines the two and is widely used as a rough marker of insulin resistance, but there is no agreed cut-off, and the number differs by about 2.3 times between mmol/L and mg/dL.

In a 28-week pilot study, middle-aged women with fatigue, stress and digestive complaints followed a functional medicine programme guided by saliva hormone and stool testing. They reported improvements on many stress, fatigue and quality-of-life measures. Salivary DHEA rose, and stool tests suggested fewer H. pylori infections. There was no control group, so the study can’t show the programme caused the improvements, or that testing first made the difference.

Statins lower the body’s own CoQ10. Whether CoQ10 supplements relieve statin-related muscle pain is unsettled. A 2020 meta-analysis of 7 placebo-controlled trials (321 patients) found no benefit for muscle symptoms or for staying on the statin (Kennedy 2020). A 2018 meta-analysis of 12 trials (575 patients) found less muscle pain, weakness and cramp, with no change in creatine kinase (Qu 2018). The difference partly reflects which trials each included.

A systematic review of 18 studies found magnesium may help subjective anxiety in people already prone to it (mildly anxious, premenstrual syndrome, high blood pressure), but rated the evidence poor and called for better trials; no study measured stress with a validated scale (Boyle 2017). A later narrative review argues that stress increases magnesium loss and low magnesium increases stress sensitivity, a “vicious circle” supported mainly by animal and mechanistic work (Pickering 2020). Three of its eight authors work for Sanofi, which sells magnesium supplements.

A meta-analysis of 17 randomised trials compared ferrous bisglycinate with other iron supplements. In pregnant women (9 trials), bisglycinate raised haemoglobin more and caused fewer gut side effects (about a third as many). In children there was no difference, and trials in other adults were too few to pool (Fischer 2023).

In 4 trials of 600 pregnant women with iron-deficiency anaemia, bovine lactoferrin raised haemoglobin at least as well as ferrous sulphate, with fewer gut side effects (Abu Hashim 2017). The trials were all in pregnancy.

A Stirling-led meta-analysis of 10 controlled studies found physical-activity programmes lowered cortisol (moderate-certainty evidence) and improved sleep quality (low certainty). Most participants were women with breast cancer; few were men and none were older adults, so the authors caution against generalising (De Nys 2022).

Reid and Fielding reviewed evidence that muscle power (force produced quickly) declines earlier and faster with age than strength, and has emerged as an important predictor of functional limitations in older adults (Reid 2012). That supports including fast, controlled movement in training as people age. It doesn’t test movement-quality coaching itself.

Across 346 early-phase cancer drug trials, treatment arms that selected patients using a biomarker had response rates of about 31%, against about 5% for arms that didn’t, and longer progression-free survival. Targeted drugs given without a biomarker did no better than standard chemotherapy. This comes from comparing trial arms rather than a randomised trial, and it is about cancer drugs designed to hit specific tumour changes.