Anne Paxton
July 2020—Gestational diabetes mellitus, if left untreated, is notoriously dangerous for mothers and their babies, making timely diagnosis critical. Yet the disease is similarly well known for being chronically under-diagnosed by laboratory testing. Could a simple change in preanalytical blood sample handling make a substantial dent in the number of undetected GDM cases?
A new study conducted in the Australian Capital Territory (ACT) in Australia suggests that it might indeed. And recognition of the role of preanalytical factors in oral glucose tolerance testing is long overdue, OGTT experts say.
The study, “Strict Preanalytical Oral Glucose Tolerance Test Blood Sample Handling Is Essential for Diagnosing Gestational Diabetes Mellitus,” was published online April 29 in Diabetes Care (https://doi.org/10.2337/dc20-0304). The study originated in the territory’s efforts to tighten up practices following the American Diabetes Association guidelines on glucose tolerance testing, says study coauthor Christopher J. Nolan, BMedSci, MBBS, PhD, FRACP endocrinologist and head of diabetes research at the Australian National University Medical School.
“The ADA guidelines say to use sodium fluoride tubes, and to take the blood sample, put it straight into ice slurry, and centrifuge within a half hour,” he says. “But they certainly weren’t doing that in the ACT.” A survey across Australia, in fact, found that most pathology laboratories weren’t following a rigorous protocol for the preanalytic collections. As the study puts it, “As there is marked heterogeneity in preanalytical OGTT blood sample handling among pathology practices in Australia, this is an issue of major importance for Australian maternity services”—and is likely not a pattern unique to Australia.
In June 2017, however, ACT Pathology changed the processing of oral glucose tolerance testing blood samples for pregnant women from a delayed to an early centrifugation protocol. That change allowed the study’s comparison of the performance of the two protocols. Between January 2015 and May 2017, OGTT samples were collected into sodium fluoride tubes and kept at room temperature until completion of the test (delayed centrifugation). Then, from June 2017 to October 2018, OGTT samples in NaF tubes were centrifuged within 10 minutes of venipuncture (early centrifugation).
A total of 7,500 women were tested with the delayed centrifugation protocol and 4,808 with the early centrifugation protocol. When the mean glucose concentrations for the fasting, one-hour, and two-hour OGTT samples were compared, they were, respectively, 0.24 mmol/L (4.32 mg/dL) (5.4 percent), 0.34 mmol/L (6.12 mg/dL) (4.9 percent), and 0.16 mmol/L (2.88 mg/dL) (2.3 percent) higher using the early centrifugation protocol. So the early centrifugation protocol increased the GDM diagnosis rate from 11.6 percent to 20.6 percent, confirming the predictions from modeling and previous small studies, the authors report.
They write, “The increase in GDM diagnosis as a result of the change [in centrifugation protocol] has had a major impact on the clinical resources required to provide GDM care.” This impact was observed in short order. “[T]he dramatic increase in the GDM diagnosis rate was evident within the first 3 months of the change to the early centrifuge preanalytical protocol.”
Dr. Nolan says he doubts Australia is alone in having a variety of types of handling of blood samples for OGTT. “I imagine [handling] varies in many countries around the world because it’s just difficult to do. For example, if you have a collection center that’s in a rural site and the samples need to be collected and sent to a central laboratory in another town or city, there’s obviously going to be delay in centrifugation. In the ACT, because it’s a fairly small jurisdiction and the public-based hospital pathology has control over all the collection centers, they were able to implement centrifuges in all the collection centers and to follow protocol closely.” But in rural areas or where there are difficulties getting samples to the laboratory—for example when blood is collected in practitioners’ consulting rooms—“there are practical issues in following the ADA guidelines.”
Much of the world, including the World Health Organization, has adopted the guidelines for GDM diagnosis approved by the International Association of the Diabetes and Pregnancy Study Groups (IADPSG). But there is more inconsistency in the U.S., Dr. Nolan says, where more than one set of criteria are acceptable and “it’s really up to clinicians to decide which one they’re going to use. However, none of the guidelines address how you should collect the blood and process the blood before it gets to the laboratory.” Pathologists, who understand the importance of preanalytical processing, “should be working a little harder among themselves,” in his view, “to reduce the variability of the preanalytic phase that’s affecting the results.”
Pathologists are also aware that the glucose levels of a blood sample, if collected into the usual sodium fluoride test tube, will fall. “Particularly if you don’t put it on ice,” Dr. Nolan notes. In fact, one reason a change in the timing of preanalytic processing of blood samples can have such a marked effect on results is that ongoing glycolysis by erythrocytes and leukocytes before centrifugation utilizes glucose, lowering its concentration.
Even putting the samples on ice still does not stop the drop in blood glucose concentration, he notes. “It only partially slows down the rate of glucose loss from the sample because the red cells and white cells keep consuming until the blood gets cold, and that causes the glucose levels to drop.” Pathologists may be less aware that glycolysis has such a large effect on the rate of diagnosis of GDM. He hopes the results of the study will convince readers that “glycolysis has a huge impact.”
Considerable evidence demonstrates that the glucose tolerance test is far better at predicting adverse pregnancy outcomes than other diagnostic tests like HbA1c, Dr. Nolan says. But the consequences of improper processing are serious. “If you lay the samples around and do a late centrifugation, a woman who does have gestational diabetes might not be diagnosed. That puts their pregnancy at further risk because they don’t know, and their clinician doesn’t know, they are at risk.” But rapid centrifugation has a downside: The risk of false-positives rises. “By using a rapid centrifugation protocol, we might actually be artificially increasing the rates of diagnosis, and then you’re treating more women who don’t need to be treated. There comes a cost if you overdiagnose,” he says.
The level of resources available to different laboratories is an important factor in setting guidelines, and the ADA recognizes that small collection centers may not always have centrifugation immediately available. In such cases, the ADA says, if placement in ice slurry and centrifugation within 30 minutes is not possible for logistical reasons, then use of citrate tubes could be considered. However, as the study’s authors point out, sodium fluoride and immediate cooling in an ice slurry does not immediately stop glycolysis. Citrate rapidly inhibits glycolysis, on the other hand, but citrate tubes may create a positive bias of 0.2 mmol/L, falsely increasing the rate of GDM diagnosis and requiring that a correction factor be applied to results.
The right way forward, Dr. Nolan expects, is to use a different type of glucose collection tube, the granulated citrate tube. The glucose level will be much more stable and results much more consistent because the citrate immediately stops the cells from using glucose without changing the glucose volume of the sample, although there is a small positive bias that might necessitate a correction of the numbers.
A strength of the study, Dr. Nolan and his coauthors say, is the large number of subjects tested before and after a well-demarcated change in OGTT protocol. And it is not hard to apply the study’s findings to other countries because the multi-ethnic community residing in the ACT, including those serving as subjects in the study, makes the findings relevant internationally. Australia generally has a multi-ethnic population, Dr. Nolan says, and in the ACT, that includes fairly high numbers of people from South Asia, Southeast Asia, the Middle East, and Northern Africa as well as the indigenous Australians, aborigines, and Pacific Islanders, who can be prone to diabetes.
One potential weakness of the study is that the two populations studied might have changed in the time interval between them. “We’re aware that with the obesity epidemic, with increasing rates of diabetes generally within the community, there’s a gradual increase in rates of diabetes, and parallel with that, gestational diabetes is increasing over time. There’s been a gradual creep up,” Dr. Nolan says.
That trend could create a risk that the study findings may relate to the epidemiology of diabetes rather than the preanalytical processing. “But what we found in our study is—just taking the rate of diagnosis three months before the change of protocol and three months after—there was a big step-up in diabetes diagnoses already. And you wouldn’t expect the population characteristics to change quickly in a way that would explain that. So we have little doubt that the problem is the preanalytical processing. The delayed centrifugation was done at different time points and that might have some small impact on the results, but I think it’s a fairly minor component of it.”
[dropcap]D[/dropcap]avid Sacks, MB, ChB, FRCPath, chief of the clinical chemistry service for the National Institutes of Health Department of Laboratory Medicine, welcomes Dr. Nolan’s study as confirmation of the critical role of preanalytical processing in diagnosing GDM. Dr. Sacks, adjunct professor of medicine at Georgetown University and clinical professor of pathology at George Washington University, is dismayed at the failure of GDM diagnosis to adequately take preanalytical processing into account. In fact, he says, “Unfortunately, diagnosis of gestational diabetes is an absolute mess.”
The root of the problem is a pivotal 1964 study by O’Sullivan and Mahan, which laid the groundwork for most of the diagnostic criteria used for GDM, he explains (Diabetes. May–June 1964;13:278–285). “They took 752 pregnant women who were asymptomatic and did three-hour glucose tolerance tests. Then they defined gestational diabetes, semi-arbitrarily, as having two of these values at more than two standard deviations above the mean.” Regrettably, these criteria, which were used to predict the development of diabetes postpartum and never looked at the effects on the pregnancy or the baby, have formed the basis for diagnosing GDM since then.
A more rigorously designed 2008 trial, the Hyperglycemia and Adverse Pregnancy Outcomes study (HAPO), set out to determine whether there was a correlation between the mother’s blood glucose and adverse pregnancy outcomes (Metzger BE, et al. N Engl J Med. 2008;358[19]:1991–2002). Looking at more than 23,000 women, the researchers tried a “one-step” oral glucose tolerance test, giving 75 grams of glucose to the women and measuring glucose at fasting and at one hour and two hours.
After studying the outcomes for the mother and baby, “they found that unfortunately, as happens in most of medicine, there was no clear cutoff. You have to use all three testing times—fasting, one-hour, and two-hour values—or you’ll miss people,” Dr. Sacks says. In saying that if only one of the three glucose tests showed an increase in the values, then you could diagnose GDM, “what happened was if you used the IADPSG criteria, you almost doubled the number of women who have gestational diabetes. That caused a big controversy.”
“The goal of the IADPSG was to get the whole world to accept the HAPO study and to get unified diagnostic criteria for GDM.” But after several consensus conferences, “that didn’t happen. As one result, in the U.S., the ADA has one set of criteria for diagnosing GDM and the obstetric community has another.”
The IADPSG criteria are evidence-based and outcome-based and show that if you have certain values there is increased risk to the mother and baby. “So therefore you have to diagnose GDM and treat these women. And other people say, ‘Well, you know, we just can’t cope with twice as many women having GDM.’ The obstetricians say that and they say, ‘We’re just going to use the old criteria which work perfectly well and we’ve been using them for 50 years.’”
It’s been a hot debate that still hasn’t been resolved because the ADA, after adopting the IADPSG criteria exclusively, walked back that move two years later and instead said several criteria could be used. So now, Dr. Sacks says, “You can have two hospitals right next to one another that have different criteria.”
The HAPO study was meticulous about sample handling, Dr. Sacks says. “They followed the criteria we proposed in laboratory guidelines for sample handling for glucose, where you collect the sample into sodium fluoride and immediately put it on ice.” Even in the presence of fluoride, glucose decreases about five to seven percent per hour at room temperature, he notes. “But anybody who tries to diagnose GDM and doesn’t follow exactly the same sample handling as HAPO will not get the same results.”
Glycolysis, then, often continues to occur at varying rates, making glucose testing results too low. “So it’s not surprising that in the study by Dr. Nolan, they found approximately a twofold increase in the prevalence of GDM,” Dr. Sacks says. “Because they handled the samples the same way it was done in the HAPO study. The large discrepancy in results was all just due to the way the tubes were handled.”
The clinical implications of correctly diagnosing GDM are huge, he says. “Modern pregnant women are very concerned about gestational diabetes. And if they’re diagnosed with it, the treatment is often insulin to manage the glucose. But some of the oral drugs normally used for diabetes can be teratogenic; they can damage the fetus. So this is an important issue.”
The delayed centrifugation described in the first part of Dr. Nolan’s study is likely what is done in many hospital and clinic settings, where the phlebotomist or nurse collects blood from the patient and keeps it there until all three samples are collected, then sends them to the lab at the same time. “So what they did in the first part of the study probably mimics what happens routinely in a lot of clinical settings in the U.S. and even the rest of the world. What they did in the second part of the study is equivalent to what was done in HAPO,” Dr. Sacks says.
As an aside, the fact that the Nolan, et al., study showed that 20 percent of the women they saw were diagnosed with GDM is remarkable, he says. “One in five pregnant women have gestational diabetes. That’s high.”
However, as Dr. Sacks and colleagues pointed out in a recent editorial in Clinical Chemistry, there is a solution to the difficulty in obtaining accurate diagnosis in GDM: obtain true measures of glucose (Bruns DE, et al. Clin Chem. 2020;66[2]:266–267). Their position is as the title says: “Diagnosis of Gestational Diabetes Mellitus Will Be Flawed Until We Can Measure Glucose.”
Studies have shown that glucose meters vary in analytical performance, they note. But the use of blood tubes containing citrate—which inhibits glycolysis completely—as well as fluoride and EDTA was found in a 2016 study to double the diagnostic sensitivity of the oral glucose tolerance test from 42 percent to 86 percent for diagnosis of GDM, with 100 percent diagnostic specificity (Daly N, et al. Clin Chem. 2016;62[6]:886–887).
Dr. Sacks and his coauthors write in their editorial: “Standardized use of CFE tubes for glucose tolerance testing for GDM would minimize or eliminate variability in diagnostic sensitivity of testing for GDM attributable to variable delays in sample processing. Standardized use of CFE tubes would also facilitate studies of the epidemiology of GDM, which are at risk of bias and virtually impossible to compare when the variable effects of glycolysis can produce a 2-fold change in the diagnostic accuracy of testing for GDM.”
The data justify concluding that U.S. laboratories need to switch to a different kind of tube, Dr. Sacks and his coauthors believe. “The tubes that have been widely used for many, many years are the sodium fluoride tubes,” Dr. Sacks says. “And they’re not perfect because the glucose metabolism in the tube continues for up to four hours. In the citrate tube, which contains citrate, fluoride, and EDTA to stop clotting, the fluoride is the same as that used in Dr. Nolan’s study, with the citrate buffer, which completely stops glycolysis.”
The problem, he says, is that although the CFE tubes are commercially available in Europe, for unknown reasons no company makes or distributes these tubes in the United States. As he and his coauthors write in their editorial, “It is unfortunate that CFE tubes are not universally commercially available for clinical use, while tubes with only NaF to inhibit glycolysis persist despite calls over the past decade to replace them with CFE tubes.” Dr. Sacks and other clinical chemists have been writing appeals in editorials and opinion pieces and he has been talking to manufacturers about the need for CFE tubes for at least 10 years, and he is perplexed that little movement has occurred. “I don’t know why they don’t make these tubes available. It’s just a baffler.”
The important conclusion to be drawn from the ACT study, Dr. Nolan says, is that “Preanalytic processing of the blood samples for glucose is critically important if you want a reliable result. And when it comes to testing for gestational diabetes mellitus, it can have a major impact on the percentage of women diagnosed. It’s a real problem that needs to be addressed.” An international consensus on recommendations, orchestrated by the professional bodies within biochemical pathology and the diabetes and pregnancy groups, will be the solution, in Dr. Nolan’s view. “We need consistency of practice around the world.”
Anne Paxton is a writer and attorney in Seattle.