Blood Tests, CT Scan Risks, and Saving Lives

Author(s): Scott Douglas Jacobsen

Publication (Outlet/Website): The Good Men Project

Publication Date (yyyy/mm/dd): 2025/06/20

Dr. Peter Bach, Chief Medical Officer of DELFI Diagnostics, discusses the urgent need to improve lung cancer screening. While low-dose CT scans reduce mortality, they are underused, with only 5–15% of eligible Americans screened. Bach explains how DELFI’s blood-based test, FirstLook Lung, analyzes cell-free DNA fragmentation to detect cancer earlier with 80% sensitivity and 99.8% negative predictive value. Bach emphasizes informed clinical decision-making and equitable access, especially for underserved populations. Widespread adoption could save up to 10,000 lives annually, making it the most impactful cancer screening strategy available.

Scott Douglas Jacobsen: So today, we’re here with Dr. Peter Bach. He is the Chief Medical Officer of Delfi Diagnostics, leading efforts to advance blood-based early detection of lung cancer. Dr. Bach is a pulmonary physician, health policy expert, and lung cancer epidemiologist. He previously served as Director of the Center for Health Policy and Outcomes at Memorial Sloan Kettering Cancer Center. His work focuses on improving cancer care through research, policy innovation, and clinical practice. He is a widely published author in leading medical journals such as The New England Journal of Medicine and JAMA and has advised national health agencies, including CMS and MedPAC. Motivated in part by the loss of his wife to breast cancer, Dr. Bach is committed to transforming cancer diagnostics and patient outcomes. Thank you very much for joining me today. So, my first question: Can you tell me a little bit about your background—an overview relevant to this particular news item around CT scans, lung cancer, and the work of Delfi?

Dr. Peter Bach: Yes, of course, Scott. The work that just came out is highly relevant to what we’re doing at Delfi. We’re focused on solving a significant public health problem: although there’s strong evidence that lung cancer screening with low-dose CT (LDCT) saves lives, it is significantly underutilized.

Studies, including large randomized trials like NLST and NELSON, have shown that annual LDCT screening for people with a significant smoking history can reduce lung cancer mortality by about 20 to 24 percent. However, adoption has been low despite being recommended in the U.S. since 2013 by the U.S. Preventive Services Task Force (USPSTF). In the U.S., screening rates hover around 5 to 15 percent among eligible adults. It’s even lower in many other countries. And even among people who do get screened, follow-up adherence—especially for annual repeat scans—is quite poor, often around 20 to 25 percent.

Part of the reason is that CT scanning, while clinically powerful, comes with challenges at scale. First, it often detects pulmonary nodules that are not cancerous. However, once these are seen, patients must undergo further imaging and, in some cases, invasive diagnostic procedures, which add anxiety, cost, and potential harm. Second, there’s a population-level concern: radiation exposure from repeated CT scans might increase cancer risk in some individuals, especially over long periods.

At DELFI, we’ve developed a blood-based test using fragmentomics—analyzing patterns of cell-free DNA fragmentation in the bloodstream. By studying the size and distribution of DNA fragments, we can detect cancer-specific signals, including those associated with lung cancer, often before symptoms appear. The goal is to use this test as an initial screen to identify who truly needs a follow-up CT scan—thereby reducing unnecessary scans, minimizing harm, and potentially improving adherence and outcomes.

And today, people aren’t getting screened as often as they should, but we do need more screening to save lives. With our approach, we can identify those patients who are most likely to have lung cancer and focus on getting them the CT scan—making each CT more likely to find actual lung cancer. That makes each unit of radiation delivered more likely to result in an early diagnosis, which is the whole purpose of screening. It also makes the nodules you find less likely to be false positives and more likely to be cancers you’re aiming to detect. That’s what we’ve been working on.

We have an AI-driven approach to genomic analysis. We’ve built our lab around a very low-cost sequencing platform, so we can offer this test at scale for just a few hundred dollars. We’ve launched this with several health systems now and are seeing improvements in screening rates.

And it all ties together—taking the best of what CT has to offer, including clinical benefits like clear visualization and next steps in diagnosis, while trying to limit radiation exposure to people who don’t actually have cancer and therefore won’t benefit from the scan.

Jacobsen: One side question comes up from that response. When you’re talking about scale in terms of screening, what was the pricing like years ago? You mentioned the test now costs a few hundred dollars—what about 10 years ago? So that we can get a sense of cost reduction over time.

Bach: You’re talking about the CT scan, the reimbursement rate has been pretty stable over the last decade, around $300. That hasn’t changed much.

Jacobsen: Okay, that’s very interesting. Now, to move into the more formal questions: What are your initial thoughts on the new study suggesting CT scans may account for up to 5% of annual cancer cases in the U.S.?

Bach: It’s frightening on its face—that the amount of imaging we’re doing in this country could be causing that much cancer.

Now, this estimate is based on lifetime exposure and population-level risk. So, while the risk to an individual patient might be quite low, say, for a specific diagnostic decision, the aggregate effect across millions of people becomes substantial. That’s the critical point. Whenever we decide to image someone, even when the risk to them is minimal and the clinical value high, it still contributes to a cumulative population-level burden.

There are valid questions about the exact size of the estimate in this study, but directionally, it’s almost certainly correct—CT scans contribute to increased cancer incidence. We should factor this into public health decisions.

I should say that CT used for lung cancer screening is explicitly not a major contributor to that 5% estimate, even if screening rates were to increase. That’s due to the limited eligible population and the relatively low radiation dose of low-dose CT (LDCT) scans. But the concern lies in the broader use of CT across medicine—in emergency departments, hospitals, and outpatient care.

Jacobsen: How should primary care physicians think about this risk-benefit balance?

Bach: I’d compare it to how we think about antibiotics for colds. The threshold for using advanced imaging, like a CT scan—should be high. Some patients benefit from imaging, and CT scans can be life-saving. But as a general rule, clinicians should ask: “Am I likely to learn something from this scan that I don’t already know?” That mindset helps reduce unnecessary tests and the downstream consequences, both clinical and societal.

Another way of saying this is: Am I likely to do something different in terms of this patient’s management once I see the scan compared to what I’m planning to do now? That’s the key question. If you want to be more technical, you can think of it as the incremental value of information.

But it’s a crucial element for any diagnostic test—especially one that carries potential harm. And, of course, there are financial costs, too. That is the basic question physicians should ask themselves, quite literally, before ordering any of these tests.

It’s also essential, Scott, not to lose sight of the fact that CT scans are necessary in many areas of patient care. I don’t want you or anyone to come away from this thinking we’re just exposing millions of people to radiation for no benefit. That’s not the case.

CT has been transformational in my medical career. For example, it has dramatically changed how we manage appendicitis. Early in my training, if you suspected appendicitis, the standard practice was to take the patient straight to surgery. A CT scan can confirm the diagnosis and avoid unnecessary surgeries. The same applies to gallbladder disease and many other conditions.

So yes, at the population level, CT scans can have harm, but they’ve also spared countless patients from invasive procedures. I could list 50 examples. CT is also critical for cancer staging, assessing disease spread, and other essential evaluations. The benefits must be carefully weighed against the risks, but there’s no question that CT is an indispensable medical tool.

Jacobsen: Are blood-based cancer screening tests reshaping diagnostic strategies and reducing radiation exposure?

Bach: Yes, especially in the context of lung cancer screening. Let’s talk about our test because it’s particularly relevant here. CT scans are currently the only screening method recommended by the USPSTF for lung cancer.

Our blood test, FirstLook Lung, is designed as an initial screening tool. And yes, a blood draw does not involve radiation. There’s a needle stick, which is mildly uncomfortable, but it’s not harmful in the way radiation exposure can be.

Our test uses fragmentomics, analyzing cell-free DNA fragmentation patterns. It has a sensitivity of about 80% overall, meaning it detects approximately 80% of lung cancers. When the result is elevated, it comes back with a clear recommendation: this person should get a follow-up CT scan.

The negative predictive value is 99.8% for patients with a non-elevated result. That means if our test does not detect a concerning fragmentation pattern, you can be 99.8% certain the patient does not have lung cancer that would be seen on a CT scan. That’s incredibly useful for informed conversations between physicians and patients, particularly those eligible for screening.

Now, armed with that result, a patient and their doctor can discuss whether they want to get a CT now or wait a year and retest.

The idea is to improve access, since blood tests are easier and more scalable than arranging CT scans—and focus resources on the subset of eligible patients most likely to benefit. Our test helps those people get the imaging they need to catch cancer early.

And importantly, our sensitivity for stage I lung cancer, which is the most curable stage, is also around 80%.

So, we can be confident that we’ll find those individuals who genuinely do have lung cancer. For everyone else, if the test comes back not elevated, we can say with high certainty that the likelihood that you have lung cancer is very, very low. Then, we can work together to make the best decision for that person—whether that means doing a CT scan now, having them come back in a year, or taking some other approach.

Jacobsen: What are the systemic drivers behind the overuse or misuse of CT scans?

Bach: I think I want to challenge the premise of the question if that’s okay.

Jacobsen: Sure.

Bach: This paper did not demonstrate widespread overuse of CT scanning. What it did show is that the collective volume of CT scans being performed is contributing to cancer incidence. And I don’t want to minimize that importance—it’s serious. It’s deeply concerning that the medical technology we use to help people may cause harm to the aggregate.

If we take the study at face value, we’re talking about a meaningful number of cancer cases caused by CT-related radiation exposure. That is not good news.

However, we must also recognize that CT scans are not being ordered “willy-nilly.” In the vast majority of cases, they’re used to evaluate legitimate signs or symptoms of disease and are highly effective in doing so. This study doesn’t tell us whether CT scans are being overused, whether they’re being used appropriately but not in the right patient groups or whether they’re being used in some combination. The study does not answer that.

That said, we must make prudent decisions with any diagnostic tool, particularly one that involves ionizing radiation. We must ask ourselves those two key questions before ordering a scan:

1. Will I learn something from this test that I don’t already know?
2. Will this test result cause me to change how I manage the patient?

Jacobsen: What are the emotional or ethical complexities physicians face with this new knowledge—that CT scans may contribute to cancer?

Bach: The most important thing, and most doctors do this already, is explaining the risks and the benefits of any medical procedure or test.

We’re accustomed to this in surgical contexts—before surgery, we’ll explain the possibility of complications like infection or bleeding and the benefits. What’s less often appreciated is that something like a CT scan, although it’s “just a test,” actually functions more like a medical procedure because of the radiation exposure. It carries both risks and benefits.

So, when physicians discuss CT scans with their patients, they should clearly explain:

• The upside is that we may gain valuable information to help diagnose or guide treatment.
• The downside is that there is a small, individual-level risk of radiation-induced cancer.

Now, for any one patient, the increase in risk is very small—almost negligible. But when you zoom out to the population level across millions, you see that it adds up. The estimate from the study suggests that up to 5% of U.S. cancer cases could be attributable to CT radiation. I think that number is somewhat overestimated—for technical reasons related to modelling assumptions—but that doesn’t change the directional truth of the finding.

Jacobsen: How does personal experience—such as losing your wife to cancer—if you’re willing to speak to it, influence your views on the urgency of developing safer diagnostics?

Bach: What I’ve gone through personally has influenced how I think about the practice of medicine in specific ways, but not on this front. I was fortunate. My wife was lucky in terms of the quality of care she received. It was both technically excellent and tender and humane.

So, no—this specific work is not personally driven in that way. What drives me in my professional life is the opportunity and importance of helping people whose lives could be saved—people who would not die of lung cancer if we screened at the levels we should be.

To give you a rough estimate, if we went from today’s lung cancer screening rates—about 10% of eligible individuals being screened—up to 80%, which is more in line with where we are for breast and colorectal cancer screening, we would save around 10,000 lives per year. That is the largest impact any cancer screening program could have, period.

And it is a substantial unmet need, particularly for an often underserved population. Smoking is more prevalent among people with lower socioeconomic status and lower educational attainment and is more common in certain minority populations.

Because of all that, this is precisely the population that could benefit the most from more accessible screening approaches, like blood tests.

That’s why I’ve devoted my career to identifying effective tools and technologies and figuring out how to get them to the people who will benefit from them.

Jacobsen: Thank you. What patient populations are more vulnerable to the harms of repeated CT imaging?

Bach: Right. I mentioned earlier that this recent study isn’t particularly relevant to low-dose CT (LDCT) lung cancer screening, and this question addresses one of the key reasons why.

When you think about how radiation causes cancer, there are two main variables:

1. How much radiation a person receives.
2. How long they live afterward—in other words, how much time there is for that radiation to result in cancer potentially.

So, those two factors combine to make younger people—and especially children—more vulnerable to radiation-induced cancer. If you’re a child and receive radiation, you simply have more years ahead of you, which increases the chance that harmful mutations will eventually lead to cancer. Contrast that with someone who’s, say, 70 years old and undergoing LDCT for lung cancer screening—their remaining lifespan is shorter, so the long-term risk is lower.

Another factor is the type and location of the CT scan. CT scans of the abdomen or pelvis often deliver higher radiation doses than scans of the lungs. That’s one reason LDCT—which uses much less radiation—is designed the way it is. It’salso why lung cancer screening with LDCT poses a relatively lower radiation risk compared to other uses of CT.

Again, while the study is important and highlights the population-level risks of CT imaging, it’s less relevant to low-dose CT screening for lung cancer. As the name implies, the “low dose” part is deliberate, and the lungs are relatively easy to image, which helps reduce the radiation exposure even further.

You don’t need that much radiation to image the lungs—because they’re mostly air. You’re essentially looking at air versus solid structures, and it doesn’t take much radiation to differentiate the two. That’s very different from scanning the abdomen, where you deal with many solid organs—like the pancreas, liver, and bowel—each with similar densities. In those cases, you need more radiation to separate one structure from another clearly.

But here’s the other important point—and I can say this because I’m in the age group: as you get older, you’re simply not around as long. So if the harm from radiation exposure manifests decades later, but you’re unlikely to be alive in 30 or 40 years, then the risk essentially disappears. That’s why, as people age—and we currently screen people up to age 80—the potential harms from radiation begin to mathematically diminish.

Jacobsen: A short follow-up: Why is the age limit set at 80 for lung cancer screening?

Bach: Great question. The reason it’s set at 80—rather than, say, 90—is based on simulation models that look at entire populations.

Here’s how they work:

• As you get older, your risk of having lung cancer increases, which argues for screening.
• But your life expectancy declines, which reduces the benefit of detecting and treating cancer.

Those two factors push in opposite directions. And age 80 is roughly where the balance still favours screening. At that point, you’re still likely to benefit—you’re likely to live long enough to receive treatment and recover.

Beyond 80, those benefits decline significantly. At 90, for instance, even if cancer is found, the likelihood of surviving long enough to benefit from treatment—surgery, recovery, follow-up-is—is low. So, national guideline bodies draw the line at 80. Of course, the real world is messier—there are many different ways to be 79 or 80—but these guidelines are based on population data and are meant to provide general direction.

Jacobsen: Last question—unless I think of a small follow-up. What healthcare innovations are helping reduce our reliance on high-radiation diagnostic tools?

Bach: Well, we’ve discussed one throughout this interview—our blood test for lung cancer. That’s a clear example of using an alternative technology that doesn’t involve radiation to identify patients most likely to benefit from a follow-up CT scan. That’s really the model: use a low-risk, scalable tool to determine who needs the higher-risk, resource-intensive follow-up.

Similar innovations exist in other areas. For example, in cardiology, we now have high-sensitivity blood tests for detecting reduced blood flow to the heart. This has helped reduce the number of patients who need to undergo cardiac catheterization, which also involves radiation.

There are likely more examples to come, but that’s the core idea: develop non-radiative, accessible diagnostics to triage and target more invasive, radiation-based tools. This improves safety, efficiency, and outcomes.

Jacobsen: Dr. Bach, I have no more questions. Thank you very much for your time and expertise today; it was a pleasure to meet you.

Bach: Thanks for your time as well, Scott. I hope you enjoy the rest of your day and your weekend.

Jacobsen: I will—I’ve got all the coffee and sunshine I need.

Bach: That’s fantastic. I’m glad to hear it.

Last updated May 3, 2025. These terms govern all In Sight Publishing content—past, present, and future—and supersede any prior notices. In Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons BY‑NC‑ND 4.0; © In Sight Publishing by Scott Douglas Jacobsen 2012–Present. All trademarks, performances, databases & branding are owned by their rights holders; no use without permission. Unauthorized copying, modification, framing or public communication is prohibited. External links are not endorsed. Cookies & tracking require consent, and data processing complies with PIPEDA & GDPR; no data from children < 13 (COPPA). Content meets WCAG 2.1 AA under the Accessible Canada Act & is preserved in open archival formats with backups. Excerpts & links require full credit & hyperlink; limited quoting under fair-dealing & fair-use. All content is informational; no liability for errors or omissions: Feedback welcome, and verified errors corrected promptly. For permissions or DMCA notices, email: scott.jacobsen2025@gmail.com. Site use is governed by BC laws; content is “as‑is,” liability limited, users indemnify us; moral, performers’ & database sui generis rights reserved.