Thursday, October 1, 2009

Pheo tests in the real world

Whenever a new test is reported, its performance is always great (why report it if not?). There are multiple reasons why the test performs well in the original reports. The diagnosis criteria are strict and clear, the tests are run by experts with strict quality control measures, the clerical errors are minimized, etc. When the test is used by more people, it usually becomes not as great. One example is the pheo test "plasma metanephrines."

In the first few studies of this test, the sensitivity (if you have pheo, the test will be positive) and specificity (if you don't have pheo, the test will be negative) both approach 100%. In a very well-designed, definitive study (published in 2002), the sensitivity and specificity are indeed close to 100% but only in people with family history of pheo. In patients without family history of pheo, the sensitivity is also nearly 100%, but the specificity is about 85%. That means, if you do not have a family history of pheo and if your doctors suspect you have pheo but you do not actually have pheo, the chance that the test will still be positive is 15% (false positive). In most cases, it is the normetanephrine that is false positive rather than metanephrine. I am sure you know what I am talking about if you have ever read a plasma metanephrines test report. The metanephrines test gives two values, metanephrine and normetanephrine.

How does the plasma metanephrines test perform in the real world? Earlier this year, a report was published comparing the performances of all 5 tests for pheo in a large US hospital. The 5 tests are plasma metanephrines, plasma catecholamines, urine metanephrines, urine catecholamines, and urine VMA. The study has quite a few limits; the major one being that not a lot of patients eventually prove to have pheo (which is also a strength because it reflects the fact that pheo is rare). With these limits, this study shows that the sensitivity and specificity of all 5 tests are about the same. The nicest thing about the study is that it gives positive predictive value (PPV) and negative predictive value (NPV) of the 5 tests in the real world.

The PPV and NPV are different from sensitivity and specificity because they depend on how the ordering physician selects patients to be tested. Let's not get entangled in technicality but the PPV of a test means that if you have a positive test for pheo, what the chance is that you actually have pheo. The NPV means that if your test result is negative, what the chance is that you actually do not have pheo. In this community-based study, the PPVs of all these tests range from 15-30%. That means if you have a pheo test at this hospital and it is positive, the chance that you indeed have pheo is about 15-30%. If the test is only moderately elevated, the PPVs are even much lower and range from 2-15%. The NPVs of all these tests are excellent: close to 100%. That means if you have a normal test result (from just any of the 5 tests) at this hospital, you are pretty sure that you do not have pheo.

Make no mistake. I do think that the plasma metanephrines is the best test for pheo. It is scientifically sound, highly reproducible, and not as interfered by medications as some other tests. I can also do clonidine suppression test based on the plasma metanephrines. I use it as the main test to diagnose pheo. It is just that the test results need to be interpreted by an expert.

Dr. Pheo