Return to CAP Home
Printable Version

  Common lab tests, uncommon problems

 

CAP Today

 

 

 

January 2010
Feature Story

Karen Lusky

Those who enjoy a good mystery can skip the detective novels and head for the clinical lab. Even seemingly routine tests can deliver a steady stream of riveting “whatdunnits.”

Take the scenario of a patient with worsening rheumatoid arthritis who was diagnosed as an inactive chronic HBV carrier based on a hepatitis B positive surface antigen test. The physician was considering doing a liver biopsy on the patient before increasing the man’s immunosuppressive therapy for arthritis—that is, until the lab discovered something that changed the picture for the patient.

Then there’s the patient diagnosed with diabetes mellitus. His A1c readings have varied significantly based on the assay used by the lab. Some of the values have been consistent with diabetes. But the patient’s latest A1c result, using a different method, is a nondiabetic value of 4.8 percent, though the fasting blood glucose test that was done simultaneously was significantly elevated. The discrepancy raises a red flag in the lab, requiring detailed investigation to sort out what’s going on.

A CAP ’09 presentation, “Pitfalls of Common Tests,” included these real-world case studies and more. As for the rest of the story for the rheumatoid arthritis patient: The physician had ordered hepatitis B surface antigen (HBsAg), HCV antibody, and liver function tests as a precaution before starting the man on immunotherapy for worsening arthritis, Bruce Beckwith, MD, chief of laboratory medicine at North­Shore Medical Center, Salem, Mass., reported in the CAP ’09 session.

The patient’s HCV antibody test was negative, and his liver function test results were normal. But the patient’s HBsAg test unexpectedly popped up with a positive result verified by repeat testing on the same sample.

The physician then ordered followup HBV testing on a second sample (HBV DNA done by PCR, hepatitis B core antibody, and hepatitis B e-antigen and e-antibody). Those tests were all negative.

Even so, the doctor diagnosed the patient as an inactive chronic HBV carrier and started the patient on an antiretroviral (lamivudine) to prevent reactivation of HBV following immunosuppressive therapy. The patient began taking methotrexate to quell his arthritis. Two months later, the clinician increased the man’s methotrexate dose and scheduled the patient for a liver biopsy, Dr. Beckwith said.

Around that time, Dr. Beckwith got on the case when a review of lab procedures revealed the lab had recently introduced new hepatitis assays. Following the change, the lab had not resumed HBsAg confirmatory testing using neutralization on low-level reactive results, including those of the arthritis patient’s. Neutralization identifies whether something in the sample other than HBV surface antigen, such as heterophile antibodies, might be causing the low reactivity.

To perform neutralization, you pretreat an aliquot from the sample with an antibody to the surface antigen, allowing it to incubate, Dr. Beckwith explained in his presentation. Then you test the sample for HBsAg, comparing the signal strength to the untreated portion of the sample. “More than a 50 percent drop in reactivity after treatment” with the antibodies confirms the positive result. “If the reactivity were due to something else . . . it should be the same,” he explains.

Dr. Beckwith notes that the lab couldn’t do neutralization on the arthritis patient’s initial sample that tested positive, as the lab no longer had the sample. But Dr. Beckwith was concerned that the followup HBV testing on a second sample was negative. (For some reason, the clinician had not ordered a repeat HBV surface antigen test on the second sample.)

Thus, Dr. Beckwith notified the patient’s physicians that the lab had not done the confirmatory testing, and suggested doing a second HBsAg on a new sample. The test result came back completely negative. The patient’s physician thus decided to stop the patient’s antiviral therapy and canceled the liver biopsy. The public health department was notified that the patient was not a chronic HBV carrier after all. And “the patient was relieved,” Dr. Beckwith reported.

The patient’s physician may have assumed the lab had performed neutralization to confirm the result, he says, adding that the lab should have done that. “Labs usually report [with the test result] that it was done. [And] whether a clinician is going to catch that it wasn’t reported depends on how lab savvy the clinician is.” In general, however, Dr. Beckwith thinks “clinicians see a positive and think it’s a positive.”

Analyzing the case requires clinicians and laboratorians to navigate the labyrinth of HBV tests, and what positive or negative results might mean in combination and within the context of a patient’s clinical status.

“HBsAg is a surface antigen that is a marker of infection that will be positive whether a person is acutely ill with hepatitis B or chronically infected. The test doesn’t distinguish between the two,” Dr. Beckwith says.

The patient’s HBV DNA by PCR was, however, negative (less than 100). But is that test more sensitive than HBsAg? It depends on whether the patient is acutely infected or a chronic carrier, Dr. Beckwith told attendees at CAP ’09. He noted that some studies predict that in acute HBV infection, you’d expect to see HBsAg be positive around the time that HBV DNA reaches 1,000 to 2,000. “However, in chronic carriers, there does not appear to be the same correlation. The ratio of circulating HBsAg to HBV DNA can vary by 1 millionfold depending upon the stage of infection.”

Weighing strongly in favor of a false-positive low-reactive HBsAg result are the patient’s other negative tests. HBV-e-antigen is a marker of viral replication, Dr. Beckwith notes. And “typically, in someone who has been infected with HBV, either the e-antigen is positive with the e-antibody being negative—or the e-antigen is negative with the e-antibody being positive.”

What does the HBV core antibody testing tell you? Alexandra Valsamakis, MD, PhD, associate professor of pathology and director of the clinical virology and molecular biology lab at Johns Hopkins, notes that if the combination assay (IgG and IgM) for the HBV core antibody is positive, but the IgM-only assay is negative, then you presume someone has core IgG and has been exposed to HBV. But the person is now fairly remote from the infection. If the combination assay is positive and the IgM positive, you can’t rule out both antibodies being present, she notes. But you do know IgM is positive and can surmise that the exposure is fairly acute.

Dr. Valsamakis says one could make the argument that you tested the patient when his HBV disease was so quiescent that everything was below the level of detection. “But you can’t fool the core IgG,” which is going to be positive, she says.

The arthritis patient’s HBsAg saga ended when he received a negative result on a new sample. But, Dr. Beckwith said, for low-reactive HBsAg results that do neutralize and don’t make sense clinically, the lab might try additional steps to see if the result is a true positive for HBsAg, as follows:

  • Recentrifuge the sample. “Some assays are sensitive to particulate matter, such as microclots,” he says.
  • Consider an alternative sample type (for example, EDTA). Dr. Beckwith says he has “seen cases using one assay where serum tubes have been weakly reactive, and . . . clearly negative using an EDTA plasma sample drawn at the same time.”
  • Run a different assay or use a blocking agent for heterophile antibodies.

Sample contamination can also cause low reactivity, an issue that can be solved by testing a new sample from the patient. Even a one percent contamination could cause a low-reactive result with the ­HBsAg test, Dr. Beckwith cautions.

If a low-reactive result persists despite the above measures, the lab and clinician face a decision tree with a number of branches, depending on the clinical situation.

For example, was the patient recently vaccinated for hepatitis B? A hepatitis B immunization, says Dr. Beckwith, can lead to surface antigen positivity for a very short period immediately after immunization; then it becomes too low to detect. While the immunization does contain surface antigen, it has no HBV DNA. “Thus, the sample should not be positive for hepatitis B DNA, if the person has never been exposed to it,” he says.

In some cases, a patient with a low-reactive HBsAg test result may have mutant strains of the virus that “escape” certain assays, Dr. Beckwith said. Antiviral therapy selects for mutant strains that have alterations in their protein sequence that elude certain assays. A person can develop mutant strains due to antiviral therapy, or he or she may become infected with a mutant strain.

The lab or clinician might consider the possibility that a patient has mutant strains when the following are seen, according to Dr. Beckwith:

  • HBsAg (+) and HBsAb (+)
  • HBsAg (-) and HBeAg or HBV DNA (+)
  • Isolated HBcAb (+)
  • Discordant HBsAg results by different assays

Sudarshan Hebbar, MD, senior medical director, Abbott Diagnostics, says while mutations occur across the entire viral genome in HBV, only specific mutations will affect assay performance. This varies by assay manufacturer, depending on which capture and detection antibodies the assay uses, he says.

Abbott Architect assays have the ability to detect the most prevalent HBV mutants, Dr. Hebbar says, which are ones at the 123 region and 145 region, with the latter being the more prevalent of the two. “These [mutations] would be detected when performing any routine testing for HBsAg, regardless of why the testing is being performed.”

Dr. Beckwith says his “take” on mutant strains is that “they aren’t terribly common in the U.S.”

A low-reactive HBsAg test can also be the tip of the iceberg in terms of identifying what’s really a roiling infection. That’s because the so-called hook effect could cause an assay to show low values when the true value is very high, Dr. Beckwith cautions. To work well, some of the assays need a certain range of antibody or antigen. “If there is too much, it overwhelms the assay,” which may show a negative or low-reactive result.

To identify the hook effect, which doesn’t affect all HBsAg assays, the sample has to be diluted, sometimes repeatedly, if large amounts of HBsAg are present, Dr. Beckwith says. (To review a diagram for hook effect, Dr. Beckwith suggests the following site: http://www.aapsj.org/articles/aapsj0902/aapsj0902017/aapsj0902017_figure2.jpg.)

In the case of the patient with arthritis, you could initially question whether the hook effect might be causing the low reactivity to HBsAg. “But the patient had no other positive antibodies,” Dr. Beckwith points out. Nor is he clinically ill, and “the hook effect is more typical in someone with very active hepatitis who is loaded with the antigen.” The clincher, however, is that HBV viral load should be positive even at very high viral levels, he says. Thus, the patient’s undetectable HBV viral load “basically rules out” hook effect.

Some of the “pearls of pathology” from his presentation, says Dr. Beckwith, include the need to evaluate positive HBsAg results in the context of other HBV test results, which the clinician did not do in this case. And keep in mind that the “strength of the signal” for an assay is important—a message he says extends beyond hepatitis B testing. “There’s a big difference between seeing the strength of the signal being just over the cutoff and four times the cutoff,” Dr. Beckwith says.

He thus believes the lab should at least report to the clinician that the result falls in a “gray area” in terms of being positive.

The key theme in a case study presented by Gary Horowitz, MD: Dig deeper when a patient’s glycated hemoglobin (hemoglobin A1c) results and blood glucose values don’t jibe. And don’t always assume the A1c is correct.

The case involved Mr. Donaldson, a patient diagnosed with diabetes mellitus, who had an A1c of 4.8 percent. A fasting blood glucose done at the same time was 138 mg/dL. The patient’s fingerstick values were, on average, 120 mg/dL.

A 4.8 percent A1c corresponds to a mean blood glucose of 91, notes Dr. Horowitz, associate professor of pathology at Harvard and director of clinical chemistry at Beth Israel Deaconess Medical Center, Boston.

Healthy individuals have a mean blood glucose of around 100, with an A1c of roughly five percent. For each one percent increase in A1c, there’s roughly a 30 mg/dL increase in mean blood glucose.

There was a key to solving the mystery of Mr. Donaldson’s discordant A1c and blood glucose values, albeit one unknown to the lab at the time. The patient had a diagnosis of hemoglobin SC disease, a form of hemoglobinopathy. People with the disease have no hemoglobin A and a shortened red blood cell life­span.

The lab found that over the previous few years, Mr. Donaldson had received A1c testing using different methods, some of which provide a glycated hemoglobin result even in the absence of hemoglobin A.

When tested using Roche’s Integra, Mr. Donaldson had received A1c readings ranging from more than six percent to 7.3 percent—values consistent with diabetes mellitus, for which the American Diabetes Association target goal is under seven percent. But Mr. Donaldson’s A1c values are inaccurate. “Integra, at least the original method, has a known limitation, yielding falsely elevated values when Hgb S or Hgb C is present,” which the insert for the assay explains, Dr. Horo­witz says.

Mr. Donaldson had also received A1c testing on the Tosoh A1c assay, which gave error messages, telling the lab not to trust the apparent A1c result because Mr. Donaldson’s specimen contained no hemoglobin A.

The patient’s clearly nondiabetic 4.8 percent A1c was obtained using the Hitachi Tina-Quant, which quantitates all glycated hemoglobins, including S and C, Dr. Horowitz said. That result was thus analytically accurate. But could it be trusted? No, Dr. Horowitz says. The value is “clinically misleading” in Mr. Donaldson’s case because he has an altered red blood cell lifespan due to his hemoglobin SC disease.

“The reference ranges [for A1c] assume a normal red blood cell lifespan [of 120 days]—and with some hemoglobinopathies and hemolytic anemias red blood cells are not around as long. [Thus], A1c is not as high as one would expect,” says Dr. Horowitz. In fact, the average red blood cell lifespan in a person with hemoglobin SC disease is 29 days, he pointed out.

Paying attention to a disconnect between the A1c readings and fingerstick blood glucose values can flag situations where the patient and clinician can’t rely on the A1c and must focus only on the patient’s blood glucose meter’s values to see how he’s doing with glucose control.

“There are times when a patient has a known condition that will affect the red blood cells’ lifespan,” says Dr. Horowitz, “but the clinician may not know that in every case.” The clinician did know that Mr. Donaldson had hemoglobin SC disease, but people in the lab did not, until they investigated the case.

Even when using the Tosoh to test for A1c, “which gave an error reading on Mr. Donaldson who has no hemoglobin A, you can get a reading on a thalassemic patient,” Dr. Horowitz says. “But because the patient’s red blood cell lifespan is abnormal, the reading would be misleading.” Again, “if you compare the A1c to the fingerstick record, you will see that they are very out of synch and know something is amiss.”

Physicians who see a puzzling disparity between a patient’s A1c and blood glucose readings might initially think horses rather than hemoglobinopathies, however. A number of relatively common conditions can affect patients’ red blood cell lifespan and thus their A1c readings. They include:

  • Any situation in which red cells aren’t lasting as long as usual (for example, hemolytic anemias). In such cases, Dr. Horowitz says, “the clinician should not even look at the person’s A1c.”
  • Significant blood loss. This results in a reduction in A1c results because the person has smaller populations of old red cells, says Darryl Erik Palmer-Toy, MD, PhD, director of chemistry at the Regional Reference Laboratories at Southern California Permanente Medical Group, North Hollywood, Calif.
  • Anemia. Patients with anemia due to nutritional deficiency may have a shift toward older red cells, which can falsely elevate the A1c result, Dr. Palmer-Toy says.

Blood transfusions, Dr. Horowitz says, can also affect the accuracy of A1c results because the patient has received blood whose A1c value is unrelated to his mean blood glucose. That effect lasts two to three months or for the life of the red blood cells.

Noncompliance with a diabetic regimen is the most common reason for a high A1c that doesn’t reflect what look to be good blood glucose readings. Dr. Horowitz says he once heard an endocrinologist drive this point home for medical residents by asking them what would cause a discrepancy between an A1c of 15 or 16 and fingerstick records that for the past month had been normal. “Everyone said a lab error. And the endocrinologist said, ‘No. This is a typical patient who doesn’t yet understand the disease and makes himself look really good the month before coming to the doctor, but the A1c tells us otherwise.’”

In another case study, Dr. Horowitz presented a real-world scenario involving Mr. Smith, a 48-year-old African-American, who was taking a statin (simva­statin) prescribed by his primary care physician for high cholesterol. The patient had other significant risk factors for cardiovascular disease, including diabetes mellitus and a strong family history of coronary artery disease.

After eight weeks of treatment with the statin, Mr. Smith’s cholesterol was markedly better and he had no symptoms. But the physician had ordered creatinine kinase to monitor Mr. Smith for statin-related myopathy. And the results came back “apparently very, very high—about 700 with an upper limit of normal [for the lab’s reference range] of about 174, roughly four times the upper limit,” Dr. Horowitz says.

The clinician called the patient and told him to stop taking the statin.

Yet fatal rhabdomyolysis is “exceptionally rare,” Dr. Horowitz said, occurring in one per 1 million prescriptions. It was “16- to 80-fold more common with cerivastatin (Baycol),” which is no longer on the market. And even with that drug, 60 percent of the cases occurred at the highest recommended dose.

In fact, the statins, it turns out, are so safe that the official clinical guidelines now don’t recommend monitoring CK unless a patient develops symptoms (muscle discomfort or weakness or brown urine), none of which Mr. Smith had. “But a lot of doctors are still doing CK,” Dr. Horowitz says.

Clinical guidelines also call for stopping statins when the CK exceeds 10 times the upper limit of normal. Yet Mr. Smith’s CK value was only four times the ULN. And “for black men, a CK result of 690 is absolutely normal,” Dr. Horowitz says. “Not only do men and woman have different upper normal limits, but with CK, the upper limit of normal differs by race as well.”

While the clinician didn’t need to order the CK on Mr. Smith or stop the treatment based on the guidelines, labs can do their part in preventing normal CK values from appearing to be very abnormal when they really aren’t.

The bottom line, Dr. Horowitz says, is that labs should do a better job establishing or verifying their reference intervals for CK. He notes, in fact, that Beth Israel Deaconess Medical Center is changing its reference range for the analyte.

“To verify a reference interval, you’d test 20 apparently healthy people, and if no more than two of the 20 were outside the [manufacturer’s] proposed reference intervals, then you’d adopt it,” Dr. Horowitz says. But suppose the group of 20 included blacks or Asians. “Chances are, you’d have more than two outside the white reference intervals.”

As an alternative to doing their own verification for reference intervals, labs can use the CAP Reference Range Service, he suggests. Labs that do so, according to CAP information on the program, “submit 20 results from ‘normal/healthy’ individuals for each analyte with information on the sex, age, and race/ethnic background of each donor.” The Reference Range Service then provides the lab with the data from other labs using the same platform.

To wrap up with the overall takeaway message from the CAP ’09 case studies, Dr. Horowitz cautions that sometimes in the “zeal” to understand the latest and greatest lab tests, “we forget to pay attention to tests that have been in our repertoire for much longer. . . .” Yet these are the tests that “tend to affect many more physicians and their patients.”


Karen Lusky is a writer in Brentwood, Tenn.
 

Related Links Related Links