HPLC testing laboratory · Davis, CaliforniaPublished pricesBook onlineinfo@openchemworks.com
Insights

Taurine and the DCM Question, in Numbers

Published November 5, 2026Updated November 19, 2026OpenChemWorks Laboratory

Updated November 19, 2026

Which studies are included

This article uses only studies that publish blood taurine as a numeric distribution (mean and SD, median and IQR, or mean and SEM) for healthy dogs split by diet category, in whole blood or plasma. Three studies meet that rule:

  1. Ontiveros et al. 2020 (PLoS ONE): 86 healthy Golden Retrievers, traditional (n=43) vs non-traditional (n=43) diets, whole blood and plasma means with SD and ranges.
  2. Adin et al. 2021 (J Vet Intern Med): 188 healthy dogs of 4 breeds, grain-free (n=26) vs grain-inclusive (n=162), whole blood and plasma medians with IQRs.
  3. Donadelli et al. 2020 (Transl Anim Sci): 8 Labrador Retrievers measured after 26 weeks on a grain-based control diet and again after 13 and 26 weeks on a grain-free diet, plasma and whole blood means with SEM. The diet contrast is sequential (same dogs, two phases), not a parallel control group, and the grain-free diet contained twice the taurine of the control (0.14% vs 0.07%), so its numbers are flagged wherever they appear.

Two well-known studies are used only as context. Kaplan et al. 2018 reports distributions by disease status (DCM cases vs healthy dogs), not diet-category means for both groups. Adin et al. 2019 (48 DCM dogs, grain-free vs grain-based) reports its taurine comparison in a figure, with no extractable numeric distribution. Because the three included studies disagree with each other, they are shown side by side and not pooled: a pooled grain-free taurine effect would be a number nobody measured.

Summary: key findings

  • The qualifying studies point in opposite directions. In Ontiveros 2020, whole blood taurine was lower on non-traditional diets: 264 ± 72.0 vs 297 ± 47.4 nmol/mL, a difference of −33 nmol/mL or −11.1% (our computation; published P = 0.01; our computed standardized effect size d = 0.54). In Adin 2021, plasma taurine was higher on grain-free diets: median 125 (IQR 101–148) vs 104 (86–123) nmol/mL, +20.2% (our computation; adjusted P = 0.02). In Donadelli 2020, taurine rose on the grain-free diet: plasma 107 → 192 nmol/mL (+79.4%) and whole blood 186 → 295 nmol/mL (+58.6%) over 26 weeks, in a diet that itself contained twice the taurine. All differences are our arithmetic on the studies' published values.
  • Whole blood and plasma do not tell the same story, even inside one study. Ontiveros: whole blood differed by diet (−11.1%, P = 0.01) while plasma did not (−3.7%, P = 0.85). Adin 2021: plasma differed after adjustment (P = 0.02) while whole blood did not (unadjusted medians 281 vs 227 nmol/mL favored grain-free, P = 0.002 unadjusted, but P = 0.99 after adjusting for age and breed). Which matrix a study leads with changes the headline. That is a measurement finding, not a diet finding, and it is the most robust result in this dataset.
  • The disease-status contrast (context, not an included study) is far larger than any diet contrast. Kaplan 2018: Golden Retrievers with taurine-deficient DCM had baseline whole blood taurine of 142.7 ± 58.2 nmol/mL (n=19) vs 279.1 ± 51.5 in 52 healthy Goldens, −48.9% (our computation). After diet change plus taurine supplementation, follow-up whole blood averaged 413.6 ± 100 nmol/mL (n=14), +189.8% vs the case baseline (different dogs in the two means; n differs, flagged). In the two parallel-group studies, diet-category differences in healthy dogs were effects of about 11% to 20% (Donadelli's sequential design, with doubled dietary taurine, showed much larger rises). The case-vs-healthy difference is a 49% effect. Confusing those two scales is how this topic gets misreported.
  • "Low" taurine was common in healthy dogs on legume-rich or grain-free diets in the Kaplan screening sample: 16 of 52 healthy Goldens (30.8%, our computation) sat at or below 250 nmol/mL whole blood, and every one of the 14 low dogs with a complete diet history was eating a legume-rich and/or grain-free diet. But 36 of 52 healthy dogs, eating varied diets, were above 250. Diet category shifted the odds; it did not determine the value.
  • The FDA investigation, as of the last public update we could verify (December 23, 2022): 524 reports (515 canine, 9 feline) covering January 1, 2014 to April 30, 2019 in the detailed dataset; more than 90% of reported products grain-free, 93% containing peas and/or lentils; FDA laboratory testing found average protein, fat, taurine, cystine, and methionine content similar between grain-free and grain-containing products; and FDA stated adverse-event counts alone do not establish a causal relationship and that it does not intend further public updates until meaningful new scientific information emerges. Later summaries put total reports received 2014–2022 at 1,382. Association in reports is the finding. Causation is not established in the sources we read, and this article does not claim it.

So what: for a pet food brand, "grain-free lowers taurine" and "grain-free is exonerated" are both stronger claims than the published distributions support. What the numbers support is testing: your diet's taurine, methionine, and cystine content, and, where the question is live-animal status, whole blood and plasma, because the two matrices have disagreed in print.

Why it matters: DCM cases are real, some are diet-associated, and some reverse with diet change and supplementation. A brand that formulates on a headline instead of on measured amino acid values is exposed in both directions: to a real nutritional miss, and to an indefensible marketing claim.

Introduction

Few questions in pet nutrition carry as much weight, or as much confusion, as whether grain-free and other non-traditional diets lower a dog's taurine status and so contribute to dilated cardiomyopathy. Headlines have run well ahead of the distributions underneath them. This article slows the question down to what was actually measured: published blood taurine values, split by diet category, in healthy dogs.

We applied a strict inclusion rule, set out at the top of this article, and only three studies met it with extractable whole blood or plasma values by diet category: the Ontiveros, Adin and Donadelli studies. Their results are presented side by side, exactly as published, with our arithmetic on their summary statistics shown as such. A disease-status study and the regulatory record are included as labelled context, not as diet evidence, and nothing across studies is pooled into a single effect.

Read what follows with two distinctions in hand. First, whole blood and plasma are different matrices and, in this dataset, they do not always agree, even inside the same study. Second, the studies report different statistics, means in some and medians in another, so each row must be read on its own terms. For a brand or a clinician, that is the usable finding: the defensible position is a measured one, based on assayed diets and on both matrices, not on a category label.

Blood taurine by diet category: three studies, no pooling

Whole blood (WB) and plasma (pl.) values. Dots show the published central value for each group.

0100200300400µmol/LOntiveros WB non-trad.mean +/- SD264Ontiveros WB trad.mean +/- SD297Ontiveros pl. non-trad.mean +/- SD102.1Ontiveros pl. trad.mean +/- SD106Adin pl. grain-freemedian (IQR)125Adin pl. grain-incl.median (IQR)104Adin WB grain-freemedian (IQR)281Adin WB grain-incl.median (IQR)227Donadelli pl. grain-freemean, SEM in source192Donadelli pl. controlmean, SEM in source107Donadelli WB grain-freemean, SEM in source295Donadelli WB controlmean, SEM in source186
Figure 1: The studies disagree in direction and the matrices disagree within studies. Whiskers are SD or IQR as labeled in each row, and the Donadelli dots are values only, with SEM reported in the source. Nothing here is pooled.

The computed dataset

Units throughout: nmol/mL, which is numerically identical to µmol/L. "Our computation" means arithmetic on the study's published summary statistics.

Table 1. The three included studies, side by side (no pooling)

Study Design / n Matrix Diet A (non-traditional / grain-free) Diet B (traditional / grain-inclusive) Difference (our computation) Published test
Ontiveros 2020 Healthy Goldens, parallel groups, 43 vs 43 Whole blood 264 ± 72.0, range 55–473 297 ± 47.4, range 216–411 −33 (−11.1%); d = 0.54 P = 0.01
Ontiveros 2020 same Plasma 102.1 ± 34.0 (n=36) 106 ± 36.2 (n=43) −3.9 (−3.7%); d = 0.11 P = 0.85 (ns)
Adin 2021 Healthy, 4 breeds, 26 vs 162 Plasma (median, IQR) 125 (101–148) 104 (86–123) +21 (+20.2%) adjusted P = 0.02
Adin 2021 same Whole blood (median, IQR) 281 (243–316) 227 (194–278) +54 (+23.8%) unadjusted unadjusted P = 0.002; adjusted P = 0.99 (ns)
Donadelli 2020 Labradors, sequential phases, n=8 Plasma (mean, SEM 11.4) 192 at week 26 on grain-free 107 at week 0 after grain-based control phase +85 (+79.4%) P = 0.0001 (time)
Donadelli 2020 same Whole blood (mean, SEM 20.2) 295 at week 26 on grain-free 186 at week 0 after grain-based control phase +109 (+58.6%) P = 0.0021 (time)

Read Table 1 the way we had to: there is no single "diet effect on taurine" in it. There are three study-specific results, in different breeds, with different diet definitions (non-traditional includes company-size criteria in Ontiveros; grain-free is an ingredient-list definition in Adin; one named commercial diet in Donadelli), and one of the three fed a test diet with double the taurine content of its control phase.

Table 2. Context (not an included study): disease-status distributions, Kaplan 2018

Group Whole blood taurine (nmol/mL) Plasma taurine (nmol/mL)
DCM + taurine deficiency, baseline (n=19 whole blood; n=7 plasma) 142.7 ± 58.2, range 48–240; 16/19 (84%) below 200 median 28, IQR 5.3–47
Healthy Goldens (n=52) 279.1 ± 51.5, range 164–382 not reported as a group mean in the text we extracted
DCM cases, follow-up after diet change + taurine (n=14) 413.6 ± 100, range 213–604 single reported value 191 (n=1)

Diet link in Kaplan: 23 of 24 cases were fed grain-free, legume-rich, or both; none of the case diets were AAFCO feeding-trial tested. Cut-offs used by the study: whole blood ≤250 and plasma ≤60 nmol/mL defined low. Reference intervals later proposed from the Ontiveros traditional-diet group: whole blood 213–377, plasma 63–194 nmol/mL. Note what that does to Kaplan's cut-off: a whole blood value of 230 is "low" by the 250 cut-off and inside the 213–377 breed reference interval. Threshold choice is part of every prevalence number in this literature.

Perspectives

Brand. What testing actually de-risks a formulation, in order of what the data can support: (1) assay the finished diet for taurine, methionine, and cystine (the FDA's own product testing found label-category averages were similar, so category tells you nothing about your SKU); (2) if you feed-trial, measure whole blood and plasma, because Table 1 shows either one alone can miss or invent a difference; (3) keep the diet history and lot records that every one of these studies needed and most case reports lacked.

Veterinary science. Association vs causation, stated plainly from the sources: FDA reports are voluntary adverse events with reporting spikes after publicity (most reports arrived after the July 2018 announcement); case series like Kaplan show reversibility in a selected, taurine-deficient, single-breed group; controlled and cross-sectional work in healthy dogs shows small, inconsistent taurine differences and, in Adin 2021, a higher cardiac troponin I in grain-free-fed healthy dogs (median hs-cTnI 0.076 vs 0.048 ng/mL, adjusted P < 0.001) alongside higher, not lower, taurine. Whatever diet-associated DCM is, the qualifying taurine dataset does not support "it is simply taurine deficiency" as the general mechanism, and Tufts' summary of the broader case population (most DCM dogs measured had normal taurine; more than 90% in their hospital series) points the same way.

Analytical. Whole blood reflects intracellular stores and moves slowly; plasma reflects recent intake and handling, and is vulnerable to artifact (platelet/leukocyte release). Ontiveros found the two correlated only moderately (Spearman r = 0.52, r² = 0.27, n=79 with both). A testing plan that specifies one matrix without saying why is an incomplete plan. Units trap, for the record: nmol/mL and µmol/L are the same number; ng/mL is not, and taurine is not reported in ng/mL in any source used here.

Regulatory. Dated status line: FDA opened the investigation in July 2018; the detailed public dataset covers reports to April 30, 2019 (524 reports); on December 23, 2022 FDA said it would not release further public updates until meaningful new scientific information emerges, and that report counts alone do not establish causation. As of our access date (November 5, 2026) we found no newer FDA update superseding that statement. We update these figures regularly.

What to do with this

If you formulate or sell a grain-free or legume-heavy diet, the defensible position is a measured one: your diet's amino acid assay on file, and a testing plan that names both blood matrices.

What we can and can't measure for this question

Taurine is not in the standard 19 amino acid panel. AA-1 (Taurine) is the full AA-1 analysis with taurine added, $150 per sample, for diets, ingredients or plasma (from 15 µL, research use only; ask us before sending whole blood). Methionine in a diet comes from AA-T ($230, hydrolysis included; methionine can partly oxidize). Cysteine and cystine need the AA-3 workup, priced after a free feasibility call. Powders and kibble that need extraction add PREP-1 at +$30 per sample. We cannot assess cardiac status or diagnose disease.

Order samples See an example report

Overnight turnaround available, with expedited service on request. Standard results in 3 to 5 business days from receipt.

Methods & Sources

Inclusion rule for Table 1: peer-reviewed study, taurine values published as a numeric distribution (mean ± SD, median + IQR, or mean + SEM) split by a diet category. Computations: percent difference = (A − B)/B × 100 on published means/medians; Cohen's d for Ontiveros from published means and SDs with pooled SD = sqrt((SD1² + SD2²)/2). We did not pool across studies (different breeds, designs, diet definitions, and summary statistics) and we did not convert medians to means. Limitations: (1) Donadelli n=8, sequential design, test diet taurine 0.14% vs control 0.07%, manufacturer-affiliated authorship (Champion Petfoods) disclosed in the paper; its rise in taurine cannot be separated from its doubled dietary taurine. (2) Adin 2021 groups are unbalanced (26 vs 162) and breed-adjusted results differ from unadjusted ones; we report both. (3) Ontiveros "non-traditional" bundles grain-free status, legume/potato position, and manufacturer size; it is not a pure grain-free contrast. (4) Diet-archetype amino acid profiles of legume-heavy formulations (planned from USDA FoodData Central ingredient data) were not computed here and are not filled with estimates.

Sources (accessed November 5, 2026): Ontiveros ES et al., PLoS ONE 2020, 15:e0233206 (PMC7228784); Adin D et al., J Vet Intern Med 2021, 35:771–779 (PMC7995416); Donadelli RA et al., Transl Anim Sci 2020, 4:txaa141 (PMC7433909); Kaplan JL et al., PLoS ONE 2018, 13:e0209112; Adin D et al., J Vet Cardiol 2019, 21:1–9 (abstract and figure legend); Freeman LM et al., JAVMA 2018, 253:1390–1394 and Tufts DCM update, November 2018; FDA, Investigation into Potential Link Between Certain Diets and Canine Dilated Cardiomyopathy, page last updated December 23, 2022 (figures as summarized in the December 2022 update coverage).

Order builder

Have samples in mind?

Pick the tests, the sample type and any preparation. The total updates as you go, and you can book in the same place. Every report includes the chromatogram and QC behind the numbers.

For the short version, see how much does amino acid testing cost: the per-sample numbers, and why quotes vary.

  • Free amino acids$150
  • Total amino acids$230
  • Creatine$120