A blood test result ten times above the normal limit for celiac antibodies predicts real intestinal damage with near-perfect accuracy — but it catches only about 1 in 6 patients who actually have that damage. The other 82% test positive for the antibody too, just not at that extreme level.
That tension — high precision, narrow reach — sits at the center of new research published in Revista de Gastroenterología de México, and it has direct implications for how celiac disease is diagnosed and when a biopsy can realistically be skipped.
What This Means for You
The blood test at the center of this study is the anti-tissue-transglutaminase IgA test — the standard first-line blood test for celiac disease. Most labs flag a result as positive once it crosses a certain threshold. But there has been growing interest, especially in pediatric gastroenterology, in using a much higher level — 10 times the upper limit of normal — as a trigger to skip the intestinal biopsy entirely. The logic: if antibody levels are that extreme, the intestinal damage is almost certainly present.
The new research confirms that logic holds when the 10x threshold fires. When patients reached it, they had confirmed intestinal damage every single time — a positive predictive value of 100%. But the study also reveals the threshold’s blind spot: most patients with real, measurable intestinal damage never reach 10x. Their levels are elevated, sometimes significantly, but not that high. Under any protocol that only acts on the 10x cutoff, those patients are overlooked.
For families navigating a new or uncertain celiac diagnosis, this matters. A child or adult can have documented gut damage and still have antibody levels that look only “moderately” elevated. The biopsy is not a formality in those cases — it may be the only way to know for certain that damage is present.
The question comes up constantly in celiac parent communities: does an elevated blood test mean the biopsy can be skipped? This research makes clear the answer depends heavily on how elevated. And for most patients, the answer is no.
Key Takeaways
- The 10x antibody threshold is accurate when it fires, but only 1 in 6 patients with intestinal damage reaches it.
- Most celiac patients with confirmed gut damage have elevated antibody levels that fall below the 10x cutoff.
- A biopsy remains essential for most patients, especially those with moderate antibody elevations.
- A lower threshold — around 3 times the normal limit — catches more patients with damage, at the cost of some false positives.
- The diagnostic gap this study identifies is not unique to the study’s Mexican population; the statistical trade-off applies wherever these thresholds are used.
The Science
Want to understand how this actually works? We’ll walk you through the technical details below and define every term. No medical degree required.
The Test and Its Threshold
Anti-tissue-transglutaminase IgA (aTG IgA) is an antibody the immune system produces in response to gluten exposure in people with celiac disease. Labs measure how high that level is relative to a reference range. The top of that range is called the upper limit of normal (ULN).
European pediatric guidelines have, in certain circumstances, allowed doctors to diagnose celiac disease without a biopsy if aTG IgA reaches ≥10x the ULN, combined with genetic markers and clinical symptoms. The threshold was designed as a high-confidence shortcut: when it fires, it should be correct.
Researchers at a major gastroenterology referral center in Mexico City tested how well that threshold performs in practice. They studied 366 patients who had been referred for suspected celiac disease or gut symptoms requiring endoscopy, with a median age of 51 years.
What the Numbers Showed
The team graded intestinal damage using the Marsh classification, a standard scale running from early inflammation (Marsh 1–2) to progressive villous atrophy — the flattening of the tiny projections (villi) lining the small intestine that absorb nutrients. Marsh 3a through 3c marks confirmed villous atrophy and the hallmark of active celiac disease.
Of the 366 patients, 53 (14.5%) received a celiac diagnosis. A total of 107 patients had Marsh 3a–3c damage. Those 107 patients had a spread of underlying diagnoses: celiac disease (45.8%), small intestinal bacterial overgrowth (23.4%), and tropical sprue (16.8%) — a reminder that other conditions can produce the same pattern of intestinal injury.
When the researchers tested the 10x ULN threshold against the full group:
- Specificity: 100% — patients without Marsh 3a–3c damage were never falsely flagged
- Sensitivity: 17.9% — only 17.9% of patients with Marsh 3a–3c damage reached the 10x level
- Positive predictive value (PPV): 100% — every patient above the threshold had confirmed villous atrophy
- Negative predictive value (NPV): 74.7% — patients below the threshold still had a 25.3% chance of harboring villous atrophy
In plain terms: the test never sounds a false alarm at 10x, but it only sounds the alarm for about 1 in 6 patients who actually have intestinal damage.
Finding a More Sensitive Cutoff
The researchers then ran a ROC (receiver operating characteristic) analysis — a statistical method for finding the threshold that best balances catching true cases (sensitivity) against minimizing false positives (specificity).
Their optimal cutoff landed at ≥3.3x the ULN. At that level:
- Sensitivity rose to 45.3%, catching nearly half of patients with villous atrophy instead of fewer than one in five
- Specificity dropped to 89.2%, meaning more patients without damage would be flagged for further investigation
That is still far from a complete picture — missing more than half of damaged patients is a significant gap. But it represents a meaningful improvement, and the researchers suggest that using a 3x threshold to trigger biopsy referrals, rather than waiting for 10x, could identify a much larger share of patients with active disease.
What This Means for Biopsy-Free Protocols
The 10x threshold was never designed as a population-wide screening tool. It was designed as a high-confidence confirmation: when it fires, skip the biopsy. This study validates that design — the PPV is 100%. But it also clarifies what the threshold cannot do: it cannot reliably identify who does have gut damage when levels are elevated but not extreme.
For pediatric diagnosis, where some guidelines allow skipping the biopsy at 10x, the takeaway is not that those guidelines are flawed. It is that children who don’t reach 10x — the large majority with real intestinal damage — still need a biopsy to confirm or rule out active disease.
This connects to a pattern we have covered before. Our earlier piece on high prevalence of ultra-short celiac disease in children with low aTG IgA levels highlighted how antibody levels at the low end can also fail to capture the full picture of intestinal involvement. Together, these studies suggest that antibody levels — whether too low, moderate, or far short of 10x — are not sufficient alone to determine whether gut damage is present.
Placing This Study in Context
This research was conducted at a tertiary referral center in Mexico City, which shapes its findings. The patient population skewed older than a typical pediatric celiac cohort, and the mix of diagnoses causing villous atrophy — including tropical sprue, uncommon in North America and Europe — reflects a specific clinical environment. The precise numbers may shift across populations.
But the underlying finding is not geographically specific. The mathematical relationship between a threshold, its sensitivity, and its specificity operates the same way regardless of where patients live. A threshold set to catch only the highest antibody levels will always miss the majority of patients with real disease — wherever they are.
Related Coverage
References
- Fernández-Ramírez A, Aguirre-Villarreal D, Rosales-Sotomayor G, et al. Predictive capacity of anti-tissue-transglutaminase IgA antibodies to identify intestinal villous atrophy in persons with celiac disease. An observational study at a referral center in Mexico City. Rev Gastroenterol Mex (Engl Ed). 2026 Jun 15:S2255-534X(26)00050-2. PubMed