Test adds twists to lung disease diagnosis

Karen Titus

November 2023—It was a mystery, wrapped less in an enigma than a few layers of bafflement, surprise, and mild irritation. Call it the Case of the Split Lung Specimens.

The first hint something was amiss came when Alain Borczuk, MD, vice chair of anatomic pathology and co-director of thoracic pathology, Northwell Health, noticed that he and his colleagues were receiving more insufficient bronchoscopy specimens than usual. “When I say ‘increasing’—we don’t get that many bronchoscopies. It’s not like colon polyps,” says Dr. Borczuk, who is also director of oncologic pathology, Northwell Health Cancer Institute. Normally they would get a handful a week, some of them straightforward cancer cases, although these additional cases were tied to noncancerous conditions.

And then the plot thickened even further, with missing pieces—literally.

Though no guideline clearly states what constitutes an adequate specimen, Dr. Borczuk says, the samples he and his colleagues were seeing fell markedly short. In looking for a disease that involves the alveoli, rather than just the airway, one would expect at least two pieces to contain alveolar lung parenchyma, he says.

Instead, they were seeing two pieces of airway wall alone or very minimal amounts of alveolar lung parenchyma. It was unusual, to say the least. “I just hadn’t been in a situation where I had to state, in so many cases, that there were limited number of alveoli.” He says he prefers not to report such cases as inadequate sampling of the lung, given its inexact definition. But faced with this uptick in limited samples, he and his colleagues found themselves regularly debating whether to use that term.

And then a breakthrough occurred: In a multidisciplinary conference, it came to light that clinicians were splitting up the samples, sending some to the lab for traditional tissue biopsy and the rest to Veracyte, to be analyzed with the company’s Envisia Genomic Classifier.

The classifier aims to identify usual interstitial pneumonia (UIP) molecular pattern. This defining morphology of idiopathic pulmonary fibrosis, a chronic and progressive interstitial lung disease, is often identified by high-resolution CT within the appropriate clinical context. In cases that are not definitive, histology can help with the diagnosis.

The Envisia test represents a different approach. The biomarker makes a binary distinction between UIP/non-UIP in transbronchial lung biopsy samples, using a 190-gene machine learning classifier (Lasky JA, et al. Ann Am Thorac Soc. 2022;19[6]:916–924; Richeldi L, et al. Am J Respir Crit Care Med. 2021;203[2]:211–220).

Dr. Borczuk recalls how he learned where the samples were disappearing to. The case in question involved an extremely small sample that was being considered for an alternative diagnosis. His pulmonologist colleague, Arunabh Talwar, MD, noted Dr. Borczuk’s report mentioned the presence of a granuloma. But the sample was so small, Dr. Borczuk conceded to Dr. Talwar, “that the only reason I mentioned one tiny granuloma was it was the only thing there I could actually comment on.”

Dr. Talwar then told him, “The Envisia test says it’s IPF.”

Dr. Borczuk was startled. “I said, ‘What Envisia test?’”

That tiny sample led to sizable conversations at Northwell. Dr. Borczuk and other experts say such conversations will need to become de rigueur elsewhere as well (if they’re not already) as pathologists help their clinical colleagues decide whether and how to pivot (if they haven’t already) to Envisia.

The emergence of the classifier has also reinvigorated concerns about long-standing limitations in diagnosing UIP and other fibrotic interstitial lung diseases. If new methods such as a genomic classifier cause frustration, so do the approaches that came into being decades ago.

“The fundamental problem,” says Jeffrey Myers, MD, “is being able to classify diffuse parenchymal lung diseases using anything short of surgical lung biopsy.” Dr. Myers is the A. James French professor of pathology and vice chair, clinical affairs and quality, Michigan Medicine.

“Everyone wants a magical test,” adds Viera Lakticova, MD, director of interventional pulmonology and bronchoscopy, Lenox Hill Hospital, and assistant professor of medicine, Division of Pulmonary, Critical Care, and Sleep Medicine, Donald and Barbara Zucker School of Medicine at Hofstra/Northwell.

Of course they do. UIP is one of the more complex areas in pulmonary pathology. It’s not limited to idiopathic pulmonary fibrosis, and cases are less common than asthma or COPD. Wedge biopsies can provide useful information, but not every patient is a good candidate.

Conventional transbronchial lung biopsies are less invasive, but not completely so, and the yield (four to six pieces, usually 2-mm to 3-mm “pinches” of tissue) can be insufficient to make the diagnosis, versus the more generous samples from surgical lung biopsy, which are typically 3 cm or 4 cm by 2 cm, from each of the three lobes on the right side. “That turns into seven or eight slides full of tissue,” says Steve Groshong, MD, PhD, chief of pathology and medical director of clinical laboratories, National Jewish Health, Denver. Cryobiopsy, a more recently developed technique, has its own limitations.

It’s not surprising, then, that the arrival of a genomic classifier has turned heads.

Dr. Groshong was involved in the BRAVE study of the test; he, along with several pulmonary and radiology colleagues at National Jewish Health, served as one of the core reading facilities of sorts to provide centralized review (Raghu G, et al. Lancet Respir Med. 2019;7[6]:487–496).

Given his involvement, he purposely recused himself from discussions about whether to use the Envisia Genomic Classifier at the institution. “But what kind of naturally evolved is that many of our ILD docs started using it.” The data they gleaned from the test, he says, matched their clinical expectations. “So they started to believe in it, in a sense.” They now use it fairly frequently, he says, though only in certain circumstances.

Dr. Groshong explains: If, for example, a patient is age 75 or older—an age at which IPF would be a reasonable diagnosis—and the CT scan shows the classic UIP pattern of honeycombing, “we usually just stop there.” No biopsy is needed.

Those who are candidates for biopsy include patients with unusual histories: too young to have IPF, for example, or a nondiagnostic CT pattern, such as honeycombing in the upper lobe rather than the lower lobe, or absence of honeycombing despite the presence of fibrosis. If the patient is young and otherwise healthy, they would be a good candidate for a wedge biopsy, says Dr. Groshong. Otherwise, a transbronchial biopsy with Envisia makes sense.

“For our docs, anyway, they always trust the wedge biopsy more,” says Dr. Groshong, “but a lot of times it’s not an option—the patient can’t tolerate a surgical procedure like that, or they cannot afford to lose more lung function in a large biopsy.”

Against that broader background, Dr. Groshong says, some pulmonologists use Envisia frequently; others, less so. “But for most of those cases, honestly, in pathology we’re never even aware they happen. We get the transbronchs just to read out for disease and make sure there aren’t granulomas and do AFB/GMS stains—that sort of thing. But the fragments they took for the classifier usually get packed up in the procedure room and shipped off, so we’re not directly aware of those until after the fact.”

No doubt that sounds familiar to Dr. Borczuk.

He was not involved in Envisia studies and doesn’t advocate for or against the test. “I’m more focused on the practical problem of integrating the test” into clinical practice.

After learning that Envisia was being used at Northwell, he and his pathologist colleagues looked into the details of the case in question. As it turns out, he says, “It didn’t really fit the indication for the test.”

The intent of transbronchial biopsies, Dr. Borczuk explains, was never to diagnose lung fibrosis; rather, it’s to identify an alternative diagnosis. If the radiology was suggestive of a fibrotic lung process, then the decision would be made to treat the patient based on the radiology, or, alternatively, if it were still ambiguous, to do some type of wedge biopsy. Definite cases of UIP require no tissue sampling.

In the category where alternative diagnosis is the primary designation for radiology, Dr. Borczuk continues, “The Envisia test is really not appropriate,” since it provides a binary answer. “But the pretest probability of the alternative diagnosis in that group is much higher, and the transbronchial biopsy is what’s going to give you [that] insight.”

That was made harder by the errant journeys the samples were taking at Northwell and beyond. If, say, six pieces of tissue were obtained by the bronchoscopy, three were sent for Envisia testing and three for pathology. “You’ve just reduced your chance of making the alternative diagnosis by half,” Dr. Borczuk says.

Dr. Borczuk told his clinical colleagues that if they were going to send transbronchial biopsy samples for Envisia testing, “I don’t want anything sent to pathology.” After all, he reminded them, “The indication for Envisia is to answer a diagnosis that I cannot answer through transbronchial biopsy. And therefore I don’t need one.”

After smoothing ruffled feathers—in part to clarify that he wasn’t criticizing bronchoscopists’ techniques per se—“We all had a good conversation that led to the desire to create an algorithm,” a four-step process for fibrotic lung disease to make clear across the system when the Envisia test is most likely to be useful.

As Dr. Borczuk explains, the first step recognizes the presence of a fibrotic lung disease that’s associated with a known clinical diagnosis, such as sarcoidosis or a drug-induced or occupational disease. In that setting, a tissue biopsy is likely not needed. “And it will be left to the clinician as to whether they want any kind of test.”

“Step two is that it’s possibly hypersensitivity pneumonitis,” he continues. This will entail a lavage, and maybe a transbronchial biopsy, but no Envisia test. “They could also opt for a surgical biopsy, but that’s a different discussion.”

In step three, the question is whether the CT shows UIP or probable UIP. Tissue sampling is not required in most cases.

Step four involves an indeterminate UIP pattern. In that setting, says Dr. Borczuk, the idiopathic pulmonary fibrosis or the UIP is only about a 50 percent pretest possibility. A surgical biopsy would be offered, but in cases where it’s considered too risky or the patient declines, then a transbronchial biopsy would be performed, with or without Envisia. If it’s felt a larger piece of tissue is needed, then a cryobiopsy would be performed. While the cryobiopsy does provide more tissue, Envisia does not accept these specimens, Dr. Borczuk notes.

The last part of this step, he says, is that if it is in fact an alternative diagnosis, and given that Envisia addresses only the UIP/non-UIP question, if UIP/IPF is less than 25 percent likely, then a transbronchial biopsy—or a cryobiopsy if possible—would be performed, but Envisia would not.

In short, he says, the algorithm suggests using Envisia only in cases where the UIP pattern is indeterminate. This has ended the problem of splitting samples.

“This narrowed the scope,” Dr. Borczuk says. Since adopting the algorithm, “We’ve been getting more cryobiopsies in the alternative diagnoses category.” If Envisia is being used in cases in which transbronchial biopsies aren’t all that helpful, “I wouldn’t know about it again—because no one’s talking to me about Envisia results.”

In that sense, he’s come full circle since he first learned of Envisia’s use at Northwell. “But I haven’t seen the problem of inadequate biopsies since.”

Dr. Lakticova, the interventional pulmonologist at Northwell, found herself beleaguered by small samples as well, though from a different angle.

Interstitial lung disease with UIP pattern is difficult to diagnose on conventional transbronchial biopsy, she notes. “And when we perform them, the amount of tissue we are getting is frequently insufficient to make the call.”

When neither cryobiopsy nor surgical lung biopsy is an option, they’ll turn to the Envisia Genomic Classifier. A positive test lets physicians know the patient has the UIP signal, which can indicate either IPF or another disease with UIP pattern, she says. “If the test is negative, it does not mean the diagnosis is not present. It just means we may be missing it.”

It’s a slender path. “I think the utility of the test is to help in the diagnosis of idiopathic pulmonary fibrosis, not to be used as a [standalone] test,” she says. Some clinicians may also use a positive result prognostically, she adds, noting that one recent study shows faster disease progression in patients with the UIP pattern noted on Envisia testing.

“The important message is that the test is an aid,” she says. “The people who are using it see it as an additional piece of information.”

“It is a useful test,” she adds. “But it is not the magic test everyone wants.”

Talk to enough experts, and it soon becomes clear their concerns lie more with misuse of the test rather than its use. This is not an everyone’s-invited type of test; think minyan, not megachurch.

In Dr. Borczuk’s mind, “There’s no question that the Envisia can help. But it helps in the cases where the certainty is on the lower side, and the ability to get tissue is limited.” If the suspicion of UIP is high, Envisia should not be done, he says. When the question of treatment becomes somewhat binary, however, and an Envisia result can move the certainty from 50 percent to 80 to 85 percent, that can give physicians the confidence to provide a treatment that in the wrong patient could prove harmful.

If most pathologists don’t have extensive experience with these cases, the same is true for most pulmonologists. The Envisia test can be an appealing option for these clinicians, Dr. Borczuk says. “They simply know there’s a test that will give them an answer that will help guide therapy in a disease state that, frankly, doesn’t have a lot of great options.”

He compares the enthusiasm he’s seen for Envisia to that for liquid biopsy. Physicians may overlook what’s lost when a tissue biopsy is replaced by a liquid one, leading to inappropriate use. That’s where pathologist input pays off.

Dr. Borczuk advises colleagues to familiarize themselves with the literature and understand the cohort of patients in the clinical studies for Envisia, as well as outcomes and how test metrics were determined. Echoing others, Dr. Borczuk says, “Clinicians have a little bit of magical thinking about how this test works”—for example, that using any type of tissue will provide an answer. “I’m not saying the studies were not done well. But it’s a little naïve to think that the type of tissue completely doesn’t matter.” He takes it a step further and calls for future molecular tests of this type to include a preanalytic component that evaluates the tissue component. “The idea that you don’t need a preanalytic component for this test to determine whether you even have lung tissue in there is, I think, ludicrous.” While the implication is there, and his clinical colleagues have told him it doesn’t matter, “I’m not sure the data entirely supports that.”

Even if clinicians aren’t asking about the test, that doesn’t mean they’re not already using it, as Dr. Borczuk learned. Pathologists can launch the conversation themselves. “It was worth it to me,” Dr. Borczuk says. “I would have never known [what was happening] if I hadn’t asked.”

He did have a clue to start, he acknowledges. Might it be helpful to nose around even if you don’t suspect a problem?

Dr. Borczuk reckons it’s worth asking. “Because there is a possibility you’re sacrificing both diagnoses—you’re not getting the benefit of the Envisia test, and you’re sacrificing the ability to obtain an alternative diagnosis. Pathologists who do have a relationship with the people who do bronchoscopy could reach out to say, ‘Are you doing this test? How are you handling the distribution of the tissue to me?’”

That’s not impolitic? “I think it’s fine,” Dr. Borczuk says. “Most pulmonologists recognize there’s an interdisciplinary component here.”

Across the country, Brandon Larsen, MD, PhD, at Mayo Clinic in Arizona, also reports seeing changes that he attributes to a growing use of Envisia.

“In our consultation practice, biopsy volumes have dramatically decreased,” says Dr. Larsen, professor of laboratory medicine and pathology and consultant, Division of Anatomic Pathology.