Should I get a baseline serum calcitonin before starting? Working the predictive value

Accepted answer

No, not with that history. The arithmetic is decisive: at the prevalence of occult medullary thyroid carcinoma in an unselected population, the great majority of raised calcitonin results are false positives, and the cascade they trigger has real costs against essentially no expected benefit. The label's non-committal wording reflects exactly this.

The predictive value arithmetic

Set up the numbers explicitly so you can vary them and see how robust the conclusion is.

Prevalence. Medullary thyroid carcinoma incidence is on the order of 1 to 2 per 100,000 per year, and it accounts for roughly 3 to 5% of thyroid malignancies. Prevalence of undiagnosed disease in an unselected adult population is higher than annual incidence but still very low; take 20 per 100,000, that is 0.02%, as a generous estimate. Including C-cell hyperplasia as a target raises the number, but hyperplasia is not itself a disease requiring intervention, so counting it inflates the apparent yield without adding value.

Test characteristics. Modern two-site immunoassays are sensitive for established medullary carcinoma — take 95%. Specificity is the problem: using a single upper reference limit without sex partitioning, 97% is a reasonable working figure, so 3% of unaffected people flag.

Per 100,000 people screened:

• Affected: 100,000 × 0.0002 = 20 people. True positives = 20 × 0.95 = 19.
• Unaffected: 100,000 − 20 = 99,980. False positives = 99,980 × 0.03 = 2,999.
• Total positives = 19 + 2,999 = 3,018.
• Positive predictive value = 19 ÷ 3,018 = 0.0063, or 0.63%.

So roughly 158 false alarms for every true case found. Put the other way: if your calcitonin comes back raised, the probability you have medullary thyroid carcinoma is under one in a hundred.

Vary the assumptions and the conclusion survives. Push specificity up to 99% and the false positives fall to 1,000, giving a PPV of 1.9% — still 52 false alarms per case. Push prevalence up tenfold, implausibly, and PPV reaches only 6% at 97% specificity. There is no plausible parameter set in an unselected population where a positive result is more likely true than false.

Note what this does not say. Calcitonin is an excellent test where it is used properly: monitoring known medullary carcinoma after thyroidectomy, and evaluating a nodule with a suggestive clinical picture. Prevalence there is orders of magnitude higher and the same test performs well. The test is not bad; the screening application is.

What raises calcitonin without cancer

The 3% false-positive rate is not assay imprecision, it is biology. Non-malignant causes include:

• Male sex. Men have higher calcitonin than women, so a non-partitioned reference limit misclassifies men preferentially.
• Chronic kidney disease — reduced clearance, substantial at low eGFR.
• Proton pump inhibitors, through hypergastrinaemia stimulating C cells. Extremely common, frequently not mentioned by the patient, and a common cause of exactly this false positive.
• Smoking; chronic autoimmune thyroiditis; hypercalcaemia and hyperparathyroidism.
• Neuroendocrine tumours elsewhere — lung, pancreas — producing calcitonin ectopically.
• Heterophilic and anti-reagent antibodies, the classic immunoassay artefact, producing spurious results that reproduce on the same platform.
• Assay differences. Reference limits are method-specific and not interchangeable.

Notice how many of those are common in the population being offered this test.

What happens after a false positive

The cost side of the ledger, and not hypothetical. The conventional pathway: repeat the calcitonin, then thyroid ultrasound, then absent an explanation a stimulation test with calcium or pentagastrin, then fine-needle aspiration of any nodule found — and nodules are present in a large fraction of adults on high-resolution ultrasound, so something will be found. Indeterminate cytology then leads to diagnostic hemithyroidectomy in a non-trivial number of cases.

So a screening test with a PPV under 1% has a realistic path to a partial thyroidectomy, permanent levothyroxine dependence, and a small operative risk to the recurrent laryngeal nerve and parathyroids. For a person with no family history and no nodule, that trade is bad in expectation by a wide margin.

Why the label is non-committal

Because the manufacturers and regulators went through the above reasoning. The labelling states, in substance, that the value of routine monitoring of serum calcitonin or of thyroid ultrasound in treated patients is uncertain, and that such monitoring may increase the risk of unnecessary procedures owing to low test specificity. That is not evasiveness; it is the arithmetic above in regulatory prose. Professional thyroid guidance has likewise declined to endorse population calcitonin screening.

A clinic recommending it either has not done the arithmetic or is selling a panel. The charitable reading is that "a baseline lets anything subsequent be interpreted" borrows the logic that genuinely applies to creatinine and ALT — and it fails here because the follow-up test is not going to be ordered either, so the baseline has nothing to be compared against.

The version where it is worth having

With a first-degree relative with medullary thyroid carcinoma or a known MEN2 syndrome the whole calculation changes — and so does the correct test. There the drug class is contraindicated and the indicated investigation is germline RET mutation analysis, not a biomarker with 97% specificity. Also worth investigating rather than screening: a palpable neck lump, persistent hoarseness, or unexplained chronic diarrhoea with flushing.

Everything above is arithmetic and label reading, not medical advice, and the decision about what to screen belongs with a clinician who knows your family history properly.

edited 2 Dec 2025 by sian_llewellyn — tightened the wording; no substantive change