About the Prague C & M Criteria for Barrett's Oesophagus
Two measurements, both taken from the top of the gastric folds: C is the proximal limit of the circumferential columnar segment and M is the top of the longest tongue, reported as C{n}M{n}. A maximal extent below 1 cm is not conventionally diagnosed as Barrett's oesophagus, because endoscopic recognition is reliable at 1 cm and above (reliability coefficient 0.72) and poor below it (0.22). From 1 cm to under 3 cm is short segment and 3 cm or more is long segment, a distinction that drives surveillance intervals. M can never be less than C, since the maximal extent includes the circumferential segment by definition.
Formula
C = distance (cm) from the top of the gastric folds to the proximal limit of the CIRCUMFERENTIAL columnar segment
M = distance (cm) from the top of the gastric folds to the top of the LONGEST tongue
Reported as C{n}M{n}. Constraint: M ≥ C always.- Top of the gastric folds
- The landmark for both measurements — the proximal cardial notch, identified with moderate insufflation. Not the squamocolumnar junction and not the diaphragmatic pinch.
- C
- The circumferential extent. Zero where the segment consists only of tongues.
- M
- The maximal extent, which includes the circumferential segment and therefore cannot be smaller than C.
- Both distances are measured from the top of the gastric folds, with MODERATE insufflation. Over-insufflating flattens the folds and moves the landmark distally, shortening both measurements.
- M can never be less than C. If it appears to be, either the values have been transposed or M was measured to the squamocolumnar junction rather than to the top of the longest tongue.
- Islands of columnar mucosa separated from the main segment are described in the report but are NOT included in M.
- A maximal extent below 1 cm is not conventionally called Barrett's oesophagus. The validation study found endoscopic recognition reliable at 1 cm and above and poor below it, which is the evidence behind the threshold.
- The classification describes extent only. Dysplasia, visible lesions and histological confirmation of intestinal metaplasia are separate and matter more for management.
Interpreting the result
Read the classification as a baseline and a category rather than as a risk figure. Below 1 cm, record the finding descriptively — an irregular Z-line, or a columnar segment under 1 cm — and do not enter the patient into surveillance, because the endoscopic recognition is unreliable at that scale and the progression risk does not justify committing someone to lifelong endoscopy. From 1 cm the diagnosis stands, and the short versus long distinction at 3 cm drives surveillance intervals, since segment length is among the strongest predictors of progression to dysplasia and adenocarcinoma. What the classification does not tell you is the part that matters most on the day: whether there is a visible lesion. A nodule, ulcer or area of irregularity changes management far more than another centimetre of segment length, and should be inspected for carefully before any biopsies are taken.
| Score | Band | What it means | Action |
|---|---|---|---|
| M < 1 cm | Below the diagnostic threshold | Endoscopic recognition is unreliable below 1 cm (reliability coefficient 0.22, against 0.72 at ≥ 1 cm) | Record descriptively as an irregular Z-line or short columnar segment; do not enter into surveillance |
| M 1 to < 3 cm | Short-segment Barrett's oesophagus | Lower risk of progression than long segments; segment length is among the strongest predictors of progression | Seattle protocol biopsies; surveillance at the longer intervals used for short segments |
| M ≥ 3 cm | Long-segment Barrett's oesophagus | Higher risk of progression to dysplasia and adenocarcinoma | Seattle protocol biopsies; shorter surveillance intervals; careful inspection for visible lesions |
| M < C | Invalid | Not a possible finding — the maximal extent includes the circumferential segment | Re-measure both from the top of the gastric folds; check the values have not been transposed |
What the Prague C & M Criteria needs (2 inputs)
- C — circumferential extent (cm)
- From the top of the gastric folds to the proximal limit of the circumferential columnar segment. Zero is a valid and common value: a segment made entirely of tongues has C0.
- M — maximal extent (cm)
- From the top of the gastric folds to the top of the longest tongue. Cannot be less than C. Separate islands of columnar mucosa are recorded descriptively in the report but are not included in M.
What it returns
- Prague classification
- Reported as C{n}M{n}, for example C2M5. This is the string that belongs in the endoscopy report and that every future endoscopy is compared against.
- Segment category
- Below the diagnostic threshold (M under 1 cm), short segment (1 to under 3 cm), or long segment (3 cm or more).
How it is calculated
The system separates two things that had previously been conflated. A Barrett's segment is rarely a neat cylinder: it usually has a circumferential portion with tongues extending proximally from it, and those two dimensions vary independently between patients and over time in the same patient. Describing the segment with a single length therefore loses information and forces the endoscopist to choose, inconsistently, between the circumferential limit and the furthest tongue. Sharma and colleagues asked a panel of expert endoscopists to score video recordings and found that both measurements could be made with good agreement provided they were anchored to a landmark that is itself reproducible — the top of the gastric folds, which is why the criteria specify moderate insufflation. The same validation work produced the finding that underpins the modern diagnostic threshold: agreement was good for segments of 1 cm or more and collapsed below that, which is why an irregular Z-line is now described rather than diagnosed.
Facts & figures
| Notation | Means |
|---|---|
| C0M2 | No circumferential segment; a 2 cm tongue |
| C2M2 | A 2 cm segment that is entirely circumferential, with no tongues |
| C2M5 | A 2 cm circumferential segment with a tongue extending to 5 cm |
| C0M0.5 | A 0.5 cm columnar segment — below the diagnostic threshold |
| C5M3 | Impossible — M cannot be less than C |
C and M are equal when the segment has no tongues, which is a valid finding rather than an error.
| Captured | Not captured |
|---|---|
| Circumferential extent | Dysplasia grade |
| Maximal extent | Presence of intestinal metaplasia on histology |
| Whether tongues are present | Visible lesions — nodules, ulcers, irregularity |
| A reproducible baseline for comparison | Hiatus hernia size |
| Islands, which are described separately |
Everything in the right-hand column matters more for management than segment length does. Prague standardises the description; it does not stratify risk on its own.
Evidence
Development and validation — Sharma et al.
2006Development of the criteria followed by validation among a panel of expert endoscopists with a special interest in Barrett's oesophagus, scoring video recordings of endoscopies to assess agreement on the circumferential and maximal extents.
Endoscopic recognition of Barrett's oesophagus of 1 cm or more achieved a reliability coefficient of 0.72, against 0.22 for segments shorter than 1 cm — the finding behind the modern diagnostic threshold. Agreement on both the C and M measurements was good for segments above that length.
Validation among trainees
2012Multicentre study assessing whether gastroenterology trainees, rather than expert endoscopists, could apply the Prague criteria reproducibly.
Confirmed that the criteria can be applied reliably outside expert centres, which is what made adoption into routine reporting practical.
BSG guidelines on Barrett's oesophagus
2014British Society of Gastroenterology guidelines on the diagnosis and management of Barrett's oesophagus.
Adopt the Prague classification as the standard for describing segment extent, and use segment length to set surveillance intervals.
ACG updated guideline
2022Updated American College of Gastroenterology guideline on the diagnosis and management of Barrett's oesophagus.
Requires at least 1 cm of columnar-lined oesophagus for a diagnosis of Barrett's, and recommends the Prague classification for describing extent.
How it compares
Prague C & M Criteria vs Los Angeles classification
Different findings in the same organ — LA grades erosive oesophagitis, Prague describes columnar-lined oesophagus, and both may be reported from one endoscopy.
The LA classification grades mucosal breaks caused by reflux; the Prague criteria describe the extent of metaplastic columnar mucosa that chronic reflux can produce. They frequently coexist, and severe oesophagitis can conceal an underlying Barrett's segment that only becomes visible once the erosions heal — which is why repeat endoscopy after healing is recommended in LA grade C and D disease.
Prague C & M Criteria vs Free-text description of segment length
Superseded. A single length loses the distinction between the circumferential segment and the tongues, which vary independently.
Describing a segment as '5 cm of Barrett's' does not say whether that is a 5 cm circumferential cylinder or a single tongue rising from a normal Z-line, and those are different findings with different biopsy requirements. The two-number notation preserves that information and makes the description reproducible between endoscopists and over time.
Prague C & M Criteria vs EREFS
Both standardise endoscopic description rather than compute risk, but for different diseases — Prague for Barrett's, EREFS for eosinophilic oesophagitis.
Neither is a risk score; both exist because free-text endoscopic description is not comparable between operators or over time. EREFS additionally sums to a total that can be tracked through treatment, whereas Prague deliberately keeps its two measurements separate because they carry different information.
Pearls & pitfalls
- Measure from the top of the gastric folds — not the squamocolumnar junction, and not the diaphragmatic pinch. Using the wrong landmark shifts both numbers.
- Use moderate insufflation. Over-insufflating flattens the gastric folds and moves the landmark distally, shortening both measurements.
- M can never be less than C. If it appears to be, the values have usually been transposed or M was measured to the squamocolumnar junction.
- C of zero is normal and common — a segment made entirely of tongues has no circumferential component.
- C and M being equal is also valid, and means the segment has no tongues.
- Islands are described in the report but are NOT included in M. Including them overstates the extent and can move a patient into a shorter surveillance interval.
- Below 1 cm, describe rather than diagnose. Calling an irregular Z-line Barrett's oesophagus commits a patient to lifelong surveillance on a finding that cannot be recognised reliably.
- The classification says nothing about dysplasia or visible lesions, and a nodule matters far more than another centimetre of length.
- A hiatus hernia makes the landmark harder to identify; take the measurement with the hernia in mind rather than measuring from the diaphragmatic impression.
Critical actions
- Record the Prague classification explicitly in the endoscopy report — it is the baseline every subsequent endoscopy is compared against.
- Inspect the segment carefully for visible lesions before taking any biopsies; a nodule or ulcer changes management more than the length does.
- Take Seattle protocol biopsies — four quadrants every 1 to 2 cm through the segment — plus targeted biopsies of any visible abnormality.
- Describe islands separately rather than folding them into M.
- Where the maximal extent is below 1 cm, record it descriptively and do not enrol the patient in surveillance.
- Note the hiatus hernia size separately, since it affects landmark identification and is relevant to management.
- Confirm intestinal metaplasia histologically before the diagnosis is treated as established.
Why this score exists
The most consequential thing the validation produced was not the notation but a number that argued against diagnosing short segments at all. When the panel's agreement on segments under 1 cm turned out to be 0.22 — barely better than chance — it made the case that an irregular Z-line cannot be recognised reliably enough to justify the consequences of calling it Barrett's oesophagus, which are lifelong surveillance endoscopies, an insurance and psychological burden, and a small procedural risk repeated indefinitely. A classification system designed to standardise description ended up defining the lower boundary of the disease. The other quiet decision is the insistence on moderate insufflation: it looks like a technical footnote and is actually what makes the landmark reproducible, since the gastric folds flatten under pressure and take the reference point with them.
About the creator
First author, 2006 development and validation study
Led the development and validation of the Prague C & M criteria within the International Working Group for the Classification of Oesophagitis.
Co-author; also a principal author of the Los Angeles classification of oesophagitis.
Co-author of the 2006 development and validation study.
Limitations
- Describes extent only, and carries no information about dysplasia, intestinal metaplasia or visible lesions — all of which matter more for management.
- Depends on correct identification of the top of the gastric folds, which is harder in the presence of a hiatus hernia and is affected by the degree of insufflation.
- Reliability falls sharply below 1 cm, which is why segments under that length are excluded from the diagnosis rather than simply reported as small.
- Excludes islands from M by convention, so a patient with extensive islands is described as having a shorter segment than the affected mucosa suggests.
- Validated largely among expert and trained endoscopists; agreement in unselected routine practice is likely to be lower.
- Segment length is only one predictor of progression, and the classification says nothing about the others — dysplasia grade, age, sex and duration of reflux.
- The short versus long boundary at 3 cm is a convention that guides surveillance intervals rather than a validated risk cut-point.
If you are the patient
Barrett's oesophagus is a change in the lining of the lower gullet, caused by long-standing acid reflux. When it is found at endoscopy, the doctor measures how far up it extends and records it in a standard way called the Prague classification. Two numbers are taken, both measured from the top of the stomach folds: C is how far up the change goes all the way around the gullet, and M is how far up the longest finger-like extension reaches. So 'C2M5' means the change goes right around for 2 cm, with one tongue reaching 5 cm. The numbers matter for two reasons. Segments shorter than 1 cm are not counted as Barrett's at all, because they cannot be measured reliably. And longer segments carry a slightly higher risk of changing over time, so they are checked more often. The measurement is written down so that the next endoscopy, perhaps years later, can be compared with this one directly.
Frequently asked questions
What are the Prague C & M criteria?#
A standardised way of describing the extent of Barrett's oesophagus using two measurements from the top of the gastric folds: C, the circumferential extent, and M, the maximal extent to the top of the longest tongue. Reported as C{n}M{n}, for example C2M5.
Why is 1 cm the diagnostic threshold?#
Because endoscopic recognition below that length is unreliable. The validation study found a reliability coefficient of 0.72 for segments of 1 cm or more and only 0.22 below it, so an irregular Z-line is described rather than diagnosed — labelling it Barrett's would commit a patient to lifelong surveillance on a finding that cannot be identified consistently.
Can M be less than C?#
No. The maximal extent includes the circumferential segment by definition, so M is always greater than or equal to C. If M appears smaller, the values have usually been transposed or M was measured to the squamocolumnar junction rather than to the top of the longest tongue.
What does C0 mean?#
That there is no circumferential columnar segment — the Barrett's consists entirely of tongues rising from the gastro-oesophageal junction. C0M2 is a common and entirely valid finding.
Are islands included in the M measurement?#
No. Separate islands of columnar mucosa are described in the endoscopy report but are not included in M. Including them overstates the extent and can shorten the surveillance interval inappropriately.
What is the difference between short-segment and long-segment Barrett's?#
A maximal extent of 1 to under 3 cm is short segment and 3 cm or more is long segment. Segment length is among the strongest predictors of progression to dysplasia and adenocarcinoma, so long segments are conventionally surveyed at shorter intervals.
Where exactly should the measurement start from?#
The top of the gastric folds — the proximal cardial notch — assessed with moderate insufflation. Not the squamocolumnar junction and not the diaphragmatic pinch. Over-insufflating flattens the folds and moves the landmark, shortening both measurements.
Does the Prague classification tell me the cancer risk?#
Only indirectly. Length is one predictor of progression, but the classification contains no information about dysplasia, intestinal metaplasia or visible lesions — and a visible nodule changes management far more than an extra centimetre of segment does.
References
Original / primary reference
Guidelines
- Shaheen NJ, Falk GW, Iyer PG, Souza RF, Yadlapati RH, Sauer BG, Wani S. Diagnosis and Management of Barrett's Esophagus: An Updated ACG Guideline. Am J Gastroenterol. 2022;117(4):559-587.
- Fitzgerald RC, di Pietro M, Ragunath K, et al. British Society of Gastroenterology guidelines on the diagnosis and management of Barrett's oesophagus. Gut. 2014;63(1):7-42.