D-dimer trifecta: clarity on units, values, and use

Dr. Kline is certainly aboard this train. In his AACC talk, he noted, “Ninety-seven percent of the people we’re testing for PE with a CT scan don’t have it. This is an astounding number.”

For the foreseeable future, however, physicians will continue to practice in a pre-post-analytic conversion world. It’s a disheveled world at best.

Spend a little time with the D-dimer literature—particularly articles proposing use of D-dimer in clinical decision rules/guidelines—and another source of confusion quickly pops up, Dr. Moser says. Sometimes the cutoff is stated clearly in terms of both unit magnitude and unit type. Sometimes it’s also clear what assay kit was used to arrive at the cutoff. Most often, though, none of this is evident. “So then laboratories are left wondering: If my clinical colleagues are calling and asking, ‘Can I use your D-dimer test in this particular clinical application?’ it’s not always clear, even if the medical director or laboratory supervisor reads the literature in question.”

She says she welcomes this proactive approach to such inquiries. On the flip side are the calls the lab receives after clinicians have ordered a D-dimer for a use they’ve read about or heard discussed at a conference. “Then they call and say, ‘Your result is goofy. Why is your D-dimer test broken?’”

She cites several articles that illustrate the problem with the literature, including one (Parvizi J, et al. J Arthroplasty. 2018;33[5]:1309–1314) that looks at using D-dimer to predict periprosthetic joint infection. The article doesn’t state the kit or kits (the article comes from several institutions) used to develop the cutoff proposed in the criteria, she notes, or whether the cutoff is in DDU or FEU. It also refers to serum D-dimer. Since D-dimer is measured on plasma, this is the wrong sample type. “This is just one example of the confusing information out there in the literature regarding D-dimer,” she says.

“This makes it hard for laboratories to advise their colleagues: ‘Yes, this works with our test’ or ‘No, it doesn’t,’” she adds.

There’s a similar dilemma with the clinical decision rule for identifying women at low risk of VTE recurrence, who might be candidates for discontinuing anticoagulant therapy. The clinical decision rule known as HERDOO2 can help identify these women, and D-dimer is one component of that rule (along with several clinical observations).

But, as Dr. Moser notes, the article describing the HERDOO2 rule validation used a single D-dimer kit to determine the D-dimer cutoff to use, and this was stated clearly in the article (Rodger MA, et al. BMJ. 2017;356:j1065). The decision rule received further attention in a subsequent analysis (Rodger MA, et al. Thromb Res. 2018;169:82–86), which explored which commercially available D-dimer assays could be used in HERDOO2 and at what cutpoint. Several assays were considered inappropriate for use in HERDOO2, according to the authors.

This may be one of those weedy patches she alluded to earlier, “but it’s clinically important,” she says. Moreover, “These are the kinds of questions that laboratory directors and laboratory supervisors get. Somebody reads these articles and says, ‘Does this work for our patients? Does this work with our test?’ And sometimes it’s hard to tell.”

Even age-adjusted D-dimer is not without problems. Reports can sometimes be confusing, so to help clinicians, Dr. Goodwin says, “Some laboratories are reporting the manufacturer’s VTE exclusion and they’re reporting out an age-adjusted VTE exclusion value.” As long as they’re using an assay that has been reported in the literature to be appropriate for using AADD, that approach is fine. But not every assay has been properly validated or has enough literature to support its use for AADD.

Laboratorians and emergency physicians are natural allies when it comes to clarifying D-dimer. Used appropriately in those patients with low or intermediate pretest probabilities, AADD can reduce the number of patients who need a CT angiogram or high-resolution CT scan to rule out PE. If a 65-year-old patient has a D-dimer value of 550 ng/mL (measured in FEU), the straight cutoff would be considered a positive result, which would warrant an imaging study, Dr. Goodwin says. Using the age-adjusted cutoff, however, the result would be 650 ng/mL FEU, “so now we’ve excluded PE, and no imaging is needed.”

Dr. Moser puts it in a zippy way: Emergency physicians are working through the differential diagnosis of “Clot or not?” Patients with high pretest probability should proceed directly to imaging. “The positive predictive value of D-dimer isn’t so great,” she says. Where it shines in the ED is in its negative predictive value. It’s especially helpful to rule out PE in elderly patients, who often have poor renal function and are best spared unnecessary radiation and contrast exposure. D-dimer should not be used to exclude patients with active malignancies, she adds.

Concurs Dr. Goodwin, “ER physicians are well aware they have to do their pretest probability—they can’t just use D-dimer to exclude all patients who walk in.”

Not everyone understands what can lead to a false elevation in D-dimer, however. Every once in a while, Dr. Goodwin gets a call from a treating physician who’s using D-dimer in a patient for whom the assay can provide no useful information—excluding a PE in a pregnant woman. It is recommended, he said, that the D-dimer not be used in the diagnostic workup of PE in pregnancy.

Such calls don’t surprise Dr. Kline, who daily sees the impact of assay variability on his colleagues.

Emergency physicians often grapple with the different thresholds used by different manufacturers, he says. And they don’t fully appreciate the regulatory requirements pathologists face when balancing industry cutoffs and the demands of CLIA regarding test validation.

The result is a cognitive dissonance of sorts, Dr. Kline says. “They understand D-dimer perfectly well, but they don’t understand why one test has this cutoff and another test has a different cutoff. Then they lose interest.”

Speaking from what is clearly an in-the-trenches perspective, Dr. Kline says his colleagues are being neither dim nor unregenerate. “But they only have so many things they can put their attention into.” When D-dimer cutoffs become too cumbersome to sort through, “They’ll say, ‘Forget it. I’ll just order a CT scan, and then I don’t even have to think about it.’”

In his AACC talk last summer and his interview with CAP TODAY, Dr. Kline noted that emergency physicians are sometimes reluctant dance partners with pretest criteria. At AACC, he said, physicians “should use pretest probability, but they don’t.”

Moreover, he added, “D-dimer has a very, very mixed reputation [with] emergency physicians because they perceive it as being positive all of the time, because of cancer, injury, infection, old age, pregnancy. And ER docs don’t like to wait for their tests. They order things in parallel because everybody is under pressure.

“We’ve got to get patients moving,” he continued. “We see 350 patients a day in the hospital where I work. Decisions have to be made fast. And you wait for the D-dimer to come back, and it can take 90 minutes, counting the blood draw, and it’s positive. And now I’ve got to order the CT scan.” That’s wasted time for physicians who think D-dimer is plagued by false-positives.

How can labs help? Dr. Kline urges pathologists to use the comments section to remind emergency physicians how to use AADD and pretest probability adjustments with the varying thresholds. “That could be a huge benefit,” he says. “I’m talking another 10 to 15 percent of patients [who] get tested for PE can be ruled out with a blood test rather than a CT scan if we use these rules.”

Like the lab, the ED is also looking at ways to fine-tune D-dimer testing. If some pathologists view a postanalytic conversion as the next step, Dr. Kline has a goal of his own. “The next frontier,” he says, “is to use pretest probability adjustment, where we double the threshold for low pretest probability.” A recent article (Kearon C, et al. N Engl J Med. 2019;381[22]:2125–2134) hints at the potential of using low clinical pretest probability and D-dimer to rule out PE. The study’s authors report that their findings are consistent with investigations that use the YEARS diagnostic algorithm (van der Pol LM, et al. N Engl J Med. 2019;380[12]:1139–1149), another approach to adjusting pretest probability.

“This is the next thing that’s going to move the needle in terms of using D-dimer,” Dr. Kline says.

Dr. Goodwin says his philosophical belief is that the laboratory needs to ally itself with clinicians. “We work hard at that because they are the end users of our results.” That includes ensuring that the data are timely and actionable.

“When age-adjusted D-dimer first came out,” Dr. Goodwin recalls, “we put a newsletter together to communicate to the clinicians. I visited the clinicians in our emergency department; I discussed with our hematology group the differences between a fibrinogen equivalent unit and a D-dimer unit and what we report in our laboratory. And I talked to them about the pros and cons of using age-adjusted D-dimer.”

Slow, antiphonal steps are rewarding, Dr. Goodwin says, allowing him to share with his clinical colleagues what he understands about the assay and its strengths and weaknesses, but also to learn and understand from them what they need. “I don’t want to be an obstructionist. I don’t want to say, ‘Oh, you cannot use it that way.’ I want to say, ‘What can we do to get the data you need to treat your patients?’”

The biggest challenge for him—apart from a recent LIS upgrade that sent every other initiative ducking for cover temporarily—is that when AADD first emerged, much of the literature looked at assays that used FEU. Since Vermont uses DDU, “There were a fair number of requests to switch our assay. I had to explain to them that it’s not just as easy as buying a different box of reagent.” His colleagues understood, and he was recruited by an ED colleague to study how well their assay performs regarding AADD—a study Dr. Goodwin embraced but which moved to the back burner because of competing issues. “I suspect it is going to be fine, but we haven’t finished the study yet.”

His situation hints at another problem with D-dimer. The guidelines are clear, says Dr. Goodwin. To use D-dimer to exclude a VTE, the standards indicate running at least 200 to 300 patients in the low- to intermediate-risk groups. “You have to follow those patients with a negative D-dimer for three months and demonstrate that the cutoff value has to have a negative predictive value of around 98 percent,” he says. “It’s pretty challenging to do that.”

As age-adjusted D-dimer has taken root, physicians have looked to meta-analyses of studies that have performed that type of rigorous evaluation. In fact, Drs. Goodwin and Moser were authors, with members of the CAP Hemostasis and Thrombosis Committee, of an article (Goodwin AJ, et al. Ann Intern Med. 2017;166[5]:361–363) that includes a table listing the various D-dimer assays (and their manufacturer, methodology, etc.) used in clinical studies of AADD cutoffs. “You’ll notice they’re all fibrinogen equivalent units,” Dr. Goodwin says. Adding up the patients studied in each of the papers, “many of the ones that use [FEUs] actually exceed the 200 patients followed for three months.

“So I think that’s where most of the practice is coming from,” Dr. Goodwin continues, “collating data from multiple papers.”

More immediately, those who use AADD have to decide who’s going to do the math. It may be simple, but it’s still easy to stumble. Some laboratories will choose to report out the age-adjusted cutoff based on a calculation made in their LIS, Dr. Moser says. Some give a more general comment or blanket statement, providing the cutoff published in the package insert and the reference interval. “Then they’ll have an interpretive statement that says, ‘For patients over 50, here’s how you calculate the age-adjusted cutoff.’ And they leave it to the receiving physician to make that calculation.”

Always, there’s worry, she says. Labs need to state clearly what they’re doing, so physicians don’t mistakenly adjust twice, like a baseball player caught unawares on camera. “I’m not sure there’s one best way to go. There’s just some different options.”

The take-home from all this, Dr. Goodwin says, is that laboratories need to know what assay they’re running. It sounds obvious, but the detail is crucial, and as Dr. Goodwin has learned from his talks with colleagues and phone calls he’s fielded, not everyone knows the basics. Labs need to know what unit they’re reporting. They need to make sure their clinical colleagues understand the cutoff for the assay. And lab and clinician alike need to know if the AADD is applicable to their kit.

Those three things are the most common challenges Dr. Goodwin sees with D-dimer testing. And if the proposed conversion happens, that would clear away a wide swath of weeds. “That would allow clinicians to stop worrying: What unit are they reporting in?”

That would then free everyone to worry about other weeds. This is D-dimer, after all.

Karen Titus is CAP TODAY contributing editor and co-managing editor.