As a kid in rural Connecticut in the early 2000s, Kathleen McWilliams was well acquainted with the danger of ticks. After days spent playing outside in the wooded areas around her house, “our home routine was you brushed your teeth, you went to the bathroom, and you did a tick check,” she said.

So when, at age 15, McWilliams suddenly spiked a 104-degree fever, her mom immediately thought Lyme disease, the tick-borne illness caused by the bacterium Borrelia burgdorferi. She requested a Lyme test. While they waited for results, McWilliams’s symptoms morphed into near-constant dizziness and achy joints, two more markers of the disease.

But, after two weeks had passed, they got surprising news: Her Lyme test came back negative.

Had it been positive, McWilliams’s doctor would have prescribed a high dose of antibiotics that would quickly clear the infection. Without that diagnosis, McWilliams’s symptoms worsened while the doctor ran a second Lyme test at her mother’s insistence. She recalls classes in which she was so dizzy she struggled to stay upright, and soccer games in which her knees and shoulders throbbed.

“I remember feeling pretty frustrated about it, because as somebody who was never sick, it was really hard to be so sick, and not have any answers,” she said.

Two weeks later, their suspicions were confirmed: McWilliams tested positive for Lyme. After a course of antibiotics, the disease and her symptoms disappeared.

Her ordeal is a common one; every year, some 300,000 people in the United States are diagnosed with Lyme disease, according to the Centers for Disease Control and Prevention, and that number is on the rise. Some people wait months or years to get a correct Lyme diagnosis. And their cases highlight a problem: Tests for Lyme in the first month of infection are frequently wrong. When diagnosed and treated early the infection is a simple one to get rid of, but left untreated it can cause a myriad of lingering symptoms, from severe arthritis to short-term memory problems.

Now, a number of research groups are working to improve Lyme tests to catch infections in the early stages. One avenue being studied by the CDC aims to create a Lyme “signature” of small molecules in the blood — an approach that, in early testing, catches a dramatically higher share of early infections.

A tricky diagnosis

Presently there’s only one Food and Drug Administration-approved laboratory test for Lyme disease: a blood test that relies on detecting antibodies, proteins the body’s immune system makes in response to the disease. But while this approach is good at diagnosing the disease at later stages, it can miss the disease early on when antibody levels are low. In the first three weeks after infection, the test only detects Lyme 29 to 40 percent of the time. (The test is 87 percent accurate once Lyme spreads to the neurological system, and 97 percent accurate for patients who develop Lyme arthritis). The CDC cautions that because the test is not likely to be positive until 4 to 6 weeks after infection, doctors who suspect Lyme based on symptoms should prescribe antibiotics even if the test is negative.

In practice, however, this doesn’t always happen. “I have to use clinical judgement quite often in early Lyme,” said Dr. Daniel Cameron, who leads a private practice specializing in Lyme in Mt. Kisco, N.Y., and has been treating Lyme for 30 years. “But there are doctors who are still waiting for that [positive] test, and if they can’t confirm it with a test, they’re reluctant to diagnose Lyme, and they don’t treat.”

He added that some doctors may hesitate to diagnose based on symptoms because they are accustomed to using highly accurate tests for other diseases, or because Lyme symptoms can look like those of multiple other illnesses, such as chronic fatigue or fibromyalgia.

In recent years, concerns about unreliable Lyme testing have even pushed state governments to get involved. In 2013, Virginia passed a Lyme Disease Testing Disclosure Act requiring physicians to inform patients that a negative Lyme test does not necessarily mean they are Lyme-free, and Maryland followed suit in 2016 with a similar law.

“Lyme disease is a serious problem here in the United States and we really need to find solutions to some of the limitations that we have, particularly in diagnosing this infection,” said Claudia Molins, a microbiologist at the CDC. “We want a test that can be used within the first two weeks of infection and that does not rely on antibody production.”

So Molins and her colleagues are focusing on metabolomics — an approach that, rather than testing directly for the immune response to the infection, instead looks for a wide spectrum of collateral damage.

Specifically they’re looking for so-called metabolite biosignatures: the litany of sugars, peptides, lipids, amino acids, fatty acids, and nucleotides normally present in the blood.

Infections like Lyme, the thinking goes, change the levels of these things — and they do so in a predictable, measurable way.

Molins and her team tested that hypothesis by tapping a unique CDC resource — freezers filled with well-characterized blood serum samples. They used serum from 89 early Lyme patients, within the first month of infection, and 50 healthy controls to develop an algorithm to detect Lyme blood signatures.

The researchers then tested that algorithm on a larger sample of serums and found that they could diagnose 88 percent of early Lyme cases, and could differentiate Lyme from other diseases (including mononucleosis and fibromyalgia) 93 percent of the time. The results were published in a 2015 study in Clinical Infectious Diseases.

Microbiologist Elli Theel, who directs the Infectious Diseases Serology Laboratory at Mayo Clinic, calls the 2015 study “incredibly promising.” “The sensitivity they showed just in early Lyme disease patients was very high, the highest I’ve actually seen,” she said.

In an upcoming publication, the researchers also showed that metabolomics can differentiate Lyme from a similar tick-borne disease called southern tick-associated rash illness (STARI). The disease causes similar symptoms as Lyme, including a bullseye rash, and occurs in overlapping geographic regions. Currently there is no laboratory test to diagnose STARI, and little is known about how the disease progresses and how to treat it, something Molins hopes will change with better testing.

Moving to clinics

Now, the researchers are trying to make metabolomics more practical in clinical settings. Their main hurdle is equipment: Laboratories like theirs use a different kind of mass spectrometer than hospitals do, so the team is currently adapting their technique for a mass spectrometry platform already used for diagnosing metabolic birth defects in many hospitals.

“Really it makes sense to try to target a platform that’s already available instead of reinventing the wheel, so to speak,” Molins explained.

They are also working on developing better biosignatures and algorithms for identifying Lyme based on new blood samples they recently collected from clinics affiliated with Colorado State University and New York Medical College. Eventually they will begin real-time testing on patients’ serum samples collected in clinics and mailed to the CDC.

“People are very scared of having Lyme disease, and with reason. And so if you can more accurately tell them if they have this infection or not, I think it’s also [good] for their own ease of mind and understanding what they’re dealing with,” Molins said. “I think we owe it to the patients to give them a proper diagnosis.”

Republished with permission from STAT. This article originally appeared on June 28, 2017