A team of researchers at Boston Children's Hospital has identified a pair of urine biomarkers — filamin C and meprin A — that could be the first to provide accurate early diagnosis of Kawasaki disease.
While the group believes the initial discovery will require more extensive validation before the two proteins can be used in clinical diagnosis, the researchers were able to test the markers prospectively in a small number of patients. The team published the results in the journal EMBO Molecular Medicine last month, showing in a cohort of 53 KD patients and 54 controls that ELISA assays for either one of the two markers were more than 90 percent accurate at distinguishing children with KD from those with other conditions that share many of the same symptoms.
Hanno Steen, one of the study's principal investigators and director of the hospital's Proteomics Center told ProteoMonitor this week that the group developed its "high-accuracy" mass spectrometry urine proteome profiling approach in an earlier study of appendicitis, published in Annals of Emergency Medicine in 2009 (PM 6/25/2009). The team decided to apply the strategy to Kawasaki disease as well, recognizing the need for an accurate early diagnostic.
Steen said that the group is working now to organize and find funding for a multi-center trial to further validate its initial findings in several hundred patients. Though the researchers believe this validation will be necessary to establish the markers as diagnostic tools, the group has already had some contact with commercial parties interested in developing a clinical-grade test, he said.
"We are talking to two companies," he said. "They are carefully optimistic based on the results, but with only  patients in the study, they would really like to see more."
Currently KD is diagnosed by ruling out other disorders using a variety of clinical measures. But this can delay definitive diagnosis. Though the disease is highly curable with early treatment using aspirin and gammaglobulin, it can potentially lead to severe cardiovascular complications if not caught early.
"If our markers can be validated, based on our preliminary data, this would be really the first true diagnostic for Kawasaki disease," Steen said.
According to Steen, the frequency of KD in the US is about 0.1 percent, but the numbers are increasing for reasons clinicians have not been able to pin down. In Japan, the disease is much more frequent, about as common as pediatric appendicitis, he said.
For the group's initial discovery and small validation, Steen and his colleagues collaborated with Susan Kim of Boston Children's to recruit patients who entered the hospital with symptoms suggesting they might have Kawasaki disease.
The group used 15 of the first specimens collected in the study — six confirmed KD cases, six confirmed non-KD subjects, and three additional samples collected from patients with KD one month after they showed a complete response to treatment — to identify candidate diagnostic markers in urine.
After fractionating proteins in 5 mL urine aliquots, the group subjected individual fractions to liquid chromatography tandem mass-spec using a nanoflow HPLC system coupled to a hybrid linear ion trap-Fourier transform ion cyclotron resonance mass spectrometer, Thermo Scientific's LTQ FT Ultra.
Overall, the group identified more than 190 proteins specific to the KD urine samples, including a variety of proteins associated with endothelial and myocardial cell injury as well as immune regulatory molecules, they reported.
According to Steen, the team chose two of these 190 proteins — meprin A and filamin C — to test further because their biological functions made them likely players in KD pathogenesis, and because each already had a commercially available ELISA.
Using these ELISAs, the group then profiled the urine of the rest of the patients recruited for the study — a total of 53 patients who were eventually diagnosed with Kawasaki, and 54 who had other mimicking diseases — blinded to the patients' final diagnosis.
Overall, the group found that urine concentrations of both markers were significantly elevated in the KD patients as compared to those with other diagnoses, even when correcting for variations in total urine protein concentrations, as well as age, sex, race, and duration of fever symptoms.
Importantly, the researchers also found that the two proteins were elevated in patients with incomplete presentations of KD — patients who met only three out of four major diagnostic criteria for the disease — as these children may be more likely to be missed by current diagnostic procedures.
Using receiver operating characteristic analysis, the group calculated the diagnostic performance of each of the two markers as an area under the curve value of .98. This beat currently used laboratory markers like blood C-reactive protein and erythrocyte sedimentation rate either alone or in combination, the researchers reported.
"With the receiver operating characteristic, basically we got in the range of 90 percent to 95 percent sensitivity and specificity," Steen said. "For many clinical tests, often you have only 80 percent – so we are quite hopeful that it is going to work out."
The team also found that filamin C and meprin A levels correlated with response to treatment, by tracking levels of the markers in the urine of five of the study participants at diagnosis, after 24 hours of treatment, and at one month after complete clinical response.
In one patient, the researchers were able to see meprin A levels that reflected an initial response to treatment and then a relapse of disease after five and a half months. In other patients who required repeat treatment, the group measured significantly higher levels of filamin C at presentation of symptoms than in other KD subjects.
To try to confirm these results, the researchers tested a second cohort of 112 retrospective serum samples collected as part of the Pediatric Heart Network study. Using ELISAs, the group found again that both markers were highly elevated in the serum of KD patients versus non-KD controls.
According to Steen, it's possible that other proteins among the 190 candidates the group identified could also be accurate diagnostic markers, and the group plans to investigate more of them. However, considering the accuracy the researchers saw using only meprin A and filamin C, there may not be a need to develop a multi-marker test.
"We were limited with manpower and money, so we went forward with only the two proteins," Steen said. "But there are some others one could look into."
"But in terms of developing a diagnostic, from a regulatory perspective, it's much better if you just have to look at one protein or two at the most," he said.
Steen said that because Kawasaki is relatively rare, the group is seeking to partner with other centers for a multi-site validation to follow up on the findings. "At this moment the most important thing is to get larger numbers," he said. "At Children's Hospital we see 40 to 60 patients a year, so we'd have to wait 10 years to get to 500 subjects working alone."
In addition, he said the group is also currently working with the hospital's Technology Innovation and Development Office to look into getting funding to develop clinical-grade ELISAs for the two markers.
According to Steen, the team is also continuing its parallel work in appendicitis, where it has already collected validation data from almost 400 subjects.
The group's additional diagnostic interests using the urinary proteome approach include developing a prognostic tool for Wilms tumor, a pediatric kidney cancer, as well as developing diagnostics that can be used to distinguish between acute inflammatory and infectious diseases.
"When you look at who are the pediatric patients in the ER, it's non-descript fever, rashes, and so on," Steen said. "So it would be good to have a markers to differentiate quickly between viral and bacterial infection or between asthma and pneumonia, for example."