The Hunt for Lung Cancer Diagnostics

Of all the inroads made in cancer diagnostics and treatments during the past several decades, precious few have involved lung cancer. The disease is the number one cancer killer in the U.S.—more than breast, prostate, and colon cancer combined—and survival rates remain abysmally low. Just 16% of newly diagnosed patients can expect to live another 5 years. At the heart of this discouraging situation lies the fact that a robust and cost-effective early-stage diagnostic test has not been identified. Although recently there has been some good news on the diagnostic imaging front in the form of low-dose computed tomography (CT) scanning, the search is intense to identify body fluid-based biomarkers that would improve on the performance of this method.

"The problem with lung cancer is that most nodules found by CT-scanning are benign. That means we're asking the medical community now to screen so many people we'll find zillions of nodules and most of them are going to be benign," explained Pierre Massion, MD, associate professor of medicine and cancer biology at Vanderbilt University in Nashville. "So now comes the question: which of those nodules are malignant? This is a huge challenge, and opportunity really, for the in vitro diagnostics field to try and find more sensitive and specific methods than what imaging has to offer us right now. This is where the medical field is really interested in moving. We really need those biomarkers."

CT Scanning: An Imperfect Marker

Any candidate biomarker will need to be evaluated in the context of findings from the National Lung Screening Trial (NLST). This National Cancer Institute-funded endeavor involving more than 53,000 patients at 33 U.S. medical centers found that low-dose CT scanning reduced lung cancer-specific and all-cause mortality in high risk individuals by 20% and 6.7%, respective-ly, versus conventional chest radiography (N Engl J Med 2011;365:395–409). Though experts hailed the NLST as the first conclusive evidence that screening for lung cancer matters, the trial also had a downside: an impressive 96.4% and 94.5% of lesions detected by CT and x-ray turned out to be false-positives, respectively. The investigators also determined that 320 individuals would need to be scanned with CT to prevent one lung cancer death.

These figures underscore the fact that while CT scanning currently is the best early detection method for lung cancer, it is decidedly imperfect. The false-positive issue troubles many experts, Massion included. In recently updated clinical practice recommendations, the National Comprehensive Cancer Network (NCCN) observed that "perhaps the most difficult aspect of lung cancer screening is addressing the moral obligation. As part of the Hippocratic oath, physicians promise to first ‘do no harm.' If lung cancer screening is not effective, then patients may be harmed from overdiagnosis, increased testing, invasive testing or procedures, and the anxiety of a potential cancer diagnosis."

Massion, who served on the NCCN committee, is on the front lines of dealing with this quandary. "Unlike many other organ systems, the lungs are really difficult to reach and at the moment the noninvasive methods are not really good enough to establish a diagnosis. If you don't operate on these people to determine for sure whether they have lung cancer, you may have missed an opportunity to treat them and cure the cancer. So it's really important to get a diagnosis because the treatment options are diametrically different," he explained. "The current dilemma in clinical practice is many patients present with nodules, and once they're told they need an invasive procedure, they're really curious about whether there's a non-invasive blood, breath, or sputum test, and right now we have to tell them no."

At What Cost?

Another major drawback of CT scanning is cost. The tally for scanning people at elevated risk of lung cancer—all 94 million current and former smokers in the U.S.—would reach nearly $30 billion, even without considering the cost and complications of following-up on the vast majority of results that would turn out to be false positives. For those reasons, influential groups such as the American Cancer Society, American Society of Clinical Oncologists, and the U.S. Preventive Services Task Force have not immediately revised guidelines to reflect the NLST results. Indeed, with the exception of the NCCN, which in February endorsed CT scanning in select patients at high risk for lung cancer, no professional organization at this time recommends lung cancer screening by any method. The NLST investigators themselves acknowledged the challenges of relying on CT-scanning alone to diagnose the disease. "Before public policy recommendations are crafted," they wrote, "the cost-effectiveness of low-dose CT screening must be rigorously analyzed. The reduction in lung cancer mortality must be weighed against the harms from positive screening results and overdiagnosis, as well as the costs." The NCCN document also highlighted that along with the major benefits of reduced lung cancer-related morbidity and mortality, CT scanning comes with a variety of risks (see Box, below). As CLN went to press, an actuarial analysis published in the influential journal, Health Affairs, concluded that lung cancer screening in high risk populations would be cost-effective, and suggested that commercial insurers should pay for this screening. How it will factor into the debate on this issue remains to be seen.

The need for more data about and a consensus on the use of CT scanning opens the door for body fluid-based biomarkers to make a prominent entrance into the lung cancer diagnostics market, according to researchers. "The cost of screening CT scans for everybody at risk for lung cancer is enormous and a significant cost burden for the healthcare system," observed Joseph Califano, MD, professor of otolaryngology-head and neck surgery at Johns Hopkins University in Baltimore. "So it still looks like there may be a role for body fluid-based diagnostics to help improve the performance of screening procedures, especially given that molecular diagnostic tests are not that expensive and will get even less expensive. A CT scan costs about $400 to $500 plus interpretation, and all the other things that come with it. Whereas a diagnostic test can come in under $100, and only requires a centralized testing facility. It's still a worthwhile and reasonable concept."

A Crowded Field of Contenders

Califano and Massion are leading two of many research efforts closing in on hoped-for body fluid-based biomarkers for early-stage lung cancer. Indeed, the field is crowded with contenders. Plasma DNA, circulating tumor DNA, microRNAs, autoantibodies, proteomics, DNA methylation, messenger RNA, and volatile organic compounds (VOCs) have all been proposed as targets in everything from blood, serum, and urine to exhaled breath, sputum, and bronchoalveolar lavage samples (see Table, below).

Of these many lines of investigation, only one has resulted so far in a commercially available test, U.K.-based Oncimmune's EarlyCDT-Lung. This test, offered for about 2 years in the company's CLIA-certified lab in Kansas City, involves a panel of seven antigens and has shown a sensitivity of 40% and specificity of 92% in detecting autoantibodies present in lung cancer. However it has had limited adoption, possibly due to the medical community's desire for further evidence of its diagnostic relevancy. Two upcoming studies could provide such data. Oncimmune in March announced a prospective trial starting later this year in Scotland that will compare EarlyCDT-Lung and low-dose CT scanning against standard care in 15,000 enrollees. A grant has just been made for a U.S.-based study that will evaluate a screening strategy combining low dose CT scanning and Early CDT-Lung testing, according to the principal investigator of that effort, James Jett, MD, an oncologist and professor of medicine at National Jewish Health in Denver.

Jett's clinic and the Cleveland Clinic also are expected this spring to start enrolling patients in a trial of a breath test being developed by Metabolomx, a California-based company focused on diagnosing diseases through breath analysis. Results from the first clinical study involving this test, which uses a colorimetric sensor assay to identify a breath signature of VOCs associated with lung cancer, showed a c-statistic of 0.811 in distinguishing lung cancer from control subjects.

Califano's team at Johns Hopkins is pursuing two lines of investigation, involving DNA methylation and microRNAs. His lab used quantitative methylation-specific PCR to identify six promising genes from a 15-gene set involved in promoter hypermethylation. In an independent set of lung cancer patients, the investigators found that one gene had 100% specificity and 35.5% sensitivity. In patients lacking this methylated gene, DCC, a logistic regression model with the five remaining genes improved sensitivity to 75% but decreased specificity to 73%.

The group's work with microRNAs revealed that two, miR-15b and miR-27b, were differentially expressed in non small cell lung cancer patients and in a training set discriminated them from healthy controls with 100% sensitivity, specificity, positive- and negative-predictive value. Further testing in lung cancer patients and healthy controls found this pair of microRNAs predicted lung cancer with 100% sensitivity and 84% specificity.

Califano is optimistic about both efforts and isn't ready to declare either the victor. "Our group and others have looked at DNA methylation as a potential test, both in sputum and blood, and it's turned out to be, depending on the marker used, a reasonable way to indicate higher risk in a subgroup of patients. In terms of how mature the testing is, methylation markers have moved further along than microRNA, but that doesn't mean one or other is going to be a better test," he said. "For whatever reason, microRNA is a little more challenging to work with in terms of batch effects, etc. That being said, microRNA is actually pretty stable, more so than standard messenger RNA, but I think the challenges are greater in terms of making sure sample collection is appropriate and robust."

A team of German researchers also has turned its attention to microRNAs after some early promising findings involving autoantibodies. The group investigated an extensive library of peptide clones, then developed antigen panels that distinguished lung cancer from normal controls with a sensitivity of 97.9% and specificity of 97%. While not abandoning these efforts, the team felt it would find more fertile ground by investigating microRNAs, according to Petra Leidinger, PhD, scientific assistant at the Saarland University Institute of Human Genetics in Homburg.

"A problem with our autoantibody research was only a minority of cancer patients have antibodies against certain proteins, therefore we chose a huge panel of antigens, 1,827. This had good sensitivity and specificity, but the smaller the panel size, the worse the accuracy of the test," she explained. "With microRNAs, you can detect all 1,500 that are known at the moment using microarray technology or sequencing."

Leidinger and her colleagues recently reported that a 14-miRNA blood signature distinguished lung cancer patients from those with chronic obstructive pulmonary disease with 91.7% sensitivity and 89.2 % specificity. The group is continuing to pursue this line of investigation.

Is Proteomics the Answer?

Massion's team is investigating a serum protein signature of seven peaks identified by matrix-assisted laser desorption ionization mass spectrometry. In a case-control study, the signature showed sensitivity of 58% and specificity of 86%. Building on this work, the team sought to validate and quantify how much the signature added to the diagnostic accuracy of clinical assessment and CT scanning in patients presenting for evaluation of lung nodules. They found that as the size of nodules increased, the proteomic signature lost its added value, suggesting it may have the most impact in evaluating indeterminate nodules. In the cohort with smaller sized nodules, the area under the receiver operating characteristic curve was 0.61 when the investigators considered clinical data and CT images only, but increased to 0.69 with use of the proteomic signature.

"I'm really proud of that paper because it outlines ways of moving biomarkers forward," said Massion. "We had a proteomic signature that was published in 2007, and we took the exact same mass spectrometry peaks, applied it to different independent cohorts, and asked whether that validation was standing the pressure of these two cohorts. The second question was whether the biomarker was adding anything to the value of chest CT scanning in these two cohorts."

Massion also is pursuing investigations involving the field of cancerization. "The idea is, the entire airway is exposed to carcinogens and tobacco smoke and the field undergoes a series of alterations and genetic and epigenetic mutations which increase the susceptibility of developing lung cancer," he explained. "We asked whether we could probe the field of cancerization and see if the proteomic signature we found in sputum can help us determine whether there is a lung cancer elsewhere in the field. It's a longer shot because you're not really diagnosing the tumor but rather, asking whether there is a surrogate signature of a cancer present in the field."

Proteomics also is the focus of a blood-based test being developed by Quest Diagnostics. The test, which relies on SomaLogic's aptamer technology, is being designed as way to whittle down the false-positives associated CT scans. "Laboratory testing to improve the reliability of a CT test could have significant clinical value, such as by potentially reducing the incidence of unnecessary surgeries or facilitating diagnosis and curative surgery in patients whose cancer may have been missed using CT scan alone," the company said in a statement provided to CLN. The test should be available in 2013.

The Crucial Validation Step

With such a broad research enterprise underway, which of these or any other approaches will prove to be the most informative and ultimately accepted in clinical practice remains very much to be seen. Aside from demonstrating adequate performance in comparison with or addition to CT scanning, any candidate biomarker will need to clear a series of other hurdles, according to experts. Key considerations include validating the proposed marker in clinically appropriate and large enough populations, establishing its cost-effectiveness, addressing practical issues about sample collection, handling, and processing, and determining the best use of the test—as part of a screening or diagnostic strategy.

"These are all promising areas of research. We just have to see how it shakes out with time. It's a horse race, if you want to use that analogy," said Jett. "I don't know which is going to turn out to be the winner, and I don't care who is the winner, I just want a winner. We need more and better markers for early detection, whatever their source."

Califano emphasized that any candidate biomarker will need to be considered in light of the NLST findings. "If you're going to show regular low-dose CT-scanning reduced mortality then all of sudden you're going to have to figure out whether in that context an adjunctive diagnostic is going to improve on that or not," he observed. "So when you're talking about what really is the next step for these diagnostic tests, they have to fit in with any accepted clinical screening guidelines. They also need to prove that they can either cull-out patients who are not at risk of developing lung cancer and thereby reserve the more expensive screening modalities like CT scanning or that they select for those who are at really significant high risk."

He and others stressed that prospective validation of any candidate markers with sufficient statistical power would be a challenge, due in part to economics. Coordinated multicenter trials might be one way to accomplish this goal. "There are a lot of people doing Phase I or II biomarker trials looking for candidates, and now we're at the point where we actually have interesting candidates that look like they may be helpful," said Califano. "So the question is, how does one go about a coordinated validation step, because it can't be done independently among all these investigators, each of whom has their own favorite marker. There has to be some way those markers are joined in a systematic fashion in a larger cohort so you get some type of meaningful data that can inform screening guidelines."

With so much excellent research underway, the sense of optimism in the field is palpable, and couldn't come a moment too soon for Jett. "I've been involved in lung cancer for 30 years—all my career—and have seen way too many people diagnosed and die. We can't stick with the paradigm of waiting until people are symptomatic. That's the whole value of these markers," he said. "I'd put this line of research in the extremely promising category. There's intense interest and something positive's going to come out within the next five years."