Cancer Rose summary, February 20, 2023
A-the interval cancers
A retrospective cohort study* published in February 2023 in "Academic Radiology" compares the characteristics of interval breast cancers, the so-called false negatives, i.e., cancers that were not detected at mammography and occur between two screening mammograms, with breast cancers detected at screening mammography.
What is interval cancer, why is it very frustrating for women participating in screening, and what are the study's findings on their characteristics?
* A type of investigation that consists of examining, based on data present in medical files or in data registers, a defined population (the cohort) and comparing a criterion or an event (here the characteristics of breast cancer) observed with one or several other groups of individuals defined according to criteria (age, living conditions, etc.)
illustration from C. Bour's book "mammo ou pas mammo?" Souccar edition
There can be two different scenarios:
1° cancer already exists and is really "missed"-
The mammographic examination is not infallible. Dense breasts are difficult to explore, and the glandular tissue is very present, resulting in a kind of opacity on mammography that prevents cancer detection. Some cancers are said to be "infiltrating" and blend in with the breast tissue. Others are atypical in shape, and still, others are completely hidden: they are not visible; the woman feels a lump one day, but the cancer is still not identifiable on mammography.
2° interval cancer
Interval cancer is cancer that was not present at the time of the mammographic examination or was only present in the form of cells. But its aggressiveness and growth are such that it develops very quickly, in a few days, weeks or months, thus in the theoretical interval between two screening mammograms, hence its name.
This situation is very frustrating for the patient, who has been told that screening is protective and saving and feels that she has "done everything right" but is not well rewarded for her diligence.
A false negative !
Results of the study
The major conclusions drawn by the authors are that interval cancers, in comparison with those detected by mammography, are, on average :
- More frequent in women with dense breasts (almost three times more)
For the authors, breast density remained significantly associated with the development of interval cancer.
When stratified by age, breast density was only significant for women over 50. This may be because dense breast tissue is more common in younger women, being present in more than 50% of women under 50 but only in less than 30% of women over 70.
- It is more advanced and has worse biological characteristics than cancers detected by mammography. In other words, screening detects mostly low-stage cancers and carcinomas in situ, many of which contribute to the pool of overdiagnosis.
Compared with screen-detected cancers, interval cancers were more often invasive than ductal carcinoma in situ (88% versus 75%, p = 0.007).
In addition, 43% (41/96) of interval cancers were primary tumours of stage 2 or higher, compared with only 12% (139/1136) of screen-detected breast cancers (p < 0.001).
Interval cancers were most often diagnosed because of symptoms and abnormalities in the breast.
- Women with a family history of breast cancer, especially first-degree (mother, sister, daughter), compared with women diagnosed with screen-detected cancer. However, family history was insignificant in multivariate analysis (a statistical method used when multiple factors potentially influence an outcome).
Conclusion of the authors
The aggressive phenotype of interval cancers may explain why they were not visible at the initial screening examination but were detectable less than a year later. These cancers are likely to be fast-growing, either new or too small to be visible at screening. The authors note that this was explicitly studied by Gilliland et al.
Furthermore, in subset analysis, interval cancers diagnosed at high-risk screening MRI were more likely to be ductal carcinoma in situ and stage 0 or 1 primary tumours, compared with symptomatic interval cancers...
For the authors, this would confirm the usefulness of screening MRI for high-risk women with high breast density, as MRI helped identify some interval cancers at an earlier stage than interval cancers identified by patients due to a symptom in the breast.
(However, it may be objected that the discovery of an earlier stage cancer in high-risk women does not tell us whether it is an interval cancer that was actually detected earlier and will thus be prevented from progressing or whether it is an intrinsically favourable cancer that would not have progressed or would have progressed little. To learn more about the problem of additional MRIs (overdiagnosis, cascades of examinations, false positives), read : https://cancer-rose.fr/en/2021/01/26/additional-mri-screening-for-women-with-extremely-dense-breast-tissue/
Cancer Rose Commentary
We repost the commentary from the excellent blog of our fellow Dr Agibus -
In his Dragiwebdo post #386, chapter 5, Dr Agibus summarizes the study's conclusion very well by recalling the so-called "barnyard" pattern, barnyard analogy breast cancer screening -
Below is what he writes:
"One paper explores mammography and interval cancers. The authors compared cancers diagnosed on screening mammograms with those diagnosed while a previous screening mammogram had been performed. They found that interval cancers were of higher stage and poorer prognosis (triple negative, adenopathy) than cancers discovered during screening. This study confirms (or at least appears to support) that screening mammography detects cancers that are not very aggressive (rabbits and turtles, sometimes too turtles, in fact). In contrast, aggressive cancers (birds) slip through the cracks and are detected on symptoms, even in the case of regular mammograms. For reminder :
In other words, aggressive cancers are intrinsically aggressive, so they are not anticipated. Those detected by repeated mammography correspond to less severe and curable cancers, with a sufficiently long residence time in the breast so that screening can detect them, but of which a large part contributes to over-diagnosis (especially the carcinoma in situ). For explanation, read: https://cancer-rose.fr/en/2020/12/17/are-small-breast-cancers-good-because-they-are-small-or-small-because-they-are-good/
Here the authors warn of unnecessary findings in routine examinations, leading to so-called "cascades of examinations".
One of the paradoxes of modern medical imaging, they say, is that the source of our greatest achievement - the ability to image the human body in ever greater detail - is also the source of one of our most significant challenges.
The success of medical imaging as a diagnostic tool has led to a dramatic increase in its use. Technological advances make it possible to acquire images at higher resolution and in more significant numbers than ever before. This has led to an increase in the detection of findings that do not appear to be related to the primary purpose of the examination and have been referred to as "incidental," This is especially true of CT and MRI scans. Many of these are harmless, but some have significant consequences for the patient's health.
Radiologists, authors say, need to become familiar with the most common incidental findings to assess their significance in each case best and to be able to recommend appropriate further investigations, when justified, because these incidental findings have implications for the patient and the service as a whole and must be thought through carefully.
Incidental findings are defined as all findings that are not directly related to the main purpose for which the imaging examination was performed, for example, the discovery of an adrenal nodule without any complaint from the patient during a CT scan or ultrasound for abdominal pain, a common and not always very specific symptom. Or the discovery of a renal nodule during a CT scan for pulmonary disease.
The development and potential widespread introduction into clinical practice of blood tests for circulating tumour DNA may add another layer of complexity.
Read here : https://cancer-rose.fr/en/2022/09/16/liquid-biopsies-the-grail/
This increased rate of detection brings with it a number of problems. The authors explain:
“Sometimes the images themselves may include features which allow us to be reasonably confident that a particular finding is either important or not - site, size, morphology, attenuation or signal characteristics may all be helpful. In many other cases there will be doubt and a decision must be made about how best to manage this uncertainty.
If it is decided that a particular lesion cannot be dismissed as irrelevant, further imaging or other more invasive tests may be recommended. The impact on the patient can range from anxiety and minor inconvenience to real harm in the event of a complication from an invasive procedure such as biopsy or endoscopy.
Much has been written about the concept of overdiagnosis – the detection and subsequent treatment of disease which if left alone would not cause problems in the patient’s lifetime. Although the term is most commonly used in relation to screening programmes, it applies equally to IFs found in symptomatic patients.”
The story of early diagnosis is seductive, but the term cancer - as it is currently used - covers many different diseases, including some indolent lesions that traditional therapeutic strategies would overtreat. (Editor's note: one reference cited is for low-grade DCIS). It is hoped that developments in artificial intelligence will help us in the future to better stratify these patients into different management strategies, some of which may involve observation rather than intervention.
For now, there is still a significant risk that the detection and reporting of an IF will result in over treatment. Aside from the impact on the individual patient, there are significant implications for radiology services, particularly in a constrained tax-funded system...
The direct cost of follow-up investigations is one consideration but an even greater risk is the potential that an increase in the number of studies performed to follow-up incidental findings will inevitably make services harder to access for other patients, some of whom may have greater need."
" Firstly we must accept that given the uncertainties inherent in radiology practice and the limitations of the tests we use, we will not always get it right.
Next we should put ourselves in a position to make the best possible assessment as to the likely relevance of each finding. We should acquaint ourselves with the appearances of the common IF in each organ as described in the articles which follow, together with the features which in each case give the best possible steer as to their likely importance.
Finally we must recognise that choosing to mention any particular finding in a radiology report is not a neutral act – it carries consequences for the patient, for the service and for other patients. For the patient, we are potentially committing them to further tests, some of which may cause concern or even real harm. For the service we are imposing a burden…”
We all have a duty and responsibility to make medical decisions about the examinations we order and perform for the patient's benefit. Not only prescribers but also radiologists must ask themselves about the scope of what they are looking for and then, for radiologists, what they are describing. How much weight and importance should be given to what they find?
Simply listing images and leaving it up to the treating physician to decide what to do with them puts the responsibility for the outcome on the prescriber alone.
Patients, too, need to be adequately informed of what systematic, routine examinations, or examinations, as we sometimes read on prescriptions, of "reassurance" may involve for their health.
Screening examinations are not infallible or harmless. They are not unstoppable shields against diseases. They can "miss" genuine lesions. They can make the patient discover unnecessary things and fall into a disease they would never have known without them.
Incidental Findings and Low-Value Care
Invited Clinical Perspective Matthew S Davenport MD-2023 Jan 11.
Departments of Radiology and Urology, Michigan Medicine, Ann Arbor MI 48108
doi: 10.2214/AJR.22.28926. Epub ahead of print. PMID: 36629303.
Detection of incidental findings in a low-risk population generally results in low-value and potentially harmful care, including paradoxically for many cancers
Incidental findings are analogous to the results of screening tests when screening is applied to unselected, low-risk patients. They generally result in low-value and potentially harmful care. Patients with incidental findings but low risk for disease are likely to experience length bias, lead-time bias, overdiagnosis, and overtreatment that create an illusion of benefit while conferring harm. This includes incidental detection of many types of cancers that, although malignant, would have been unlikely to affect a patient’s health had the cancer not been detected. Detection of some incidental findings can create high-value care, but most do not, and differentiation is often unclear at the time of identification. Higher patient- and disease-related risk increase the likelihood an incidental finding is important. Clinical guidelines for incidental findings should more deeply integrate patient risk factors and disease aggressiveness to inform management. However, lack of outcome and cost-effectiveness data lead to reflexive management strategies that create low-value, expensive, potentially harmful care.
Radiology needs outcome and cost-effectiveness data to inform its management recommendations for incidental findings.
So, What Should We Do?
It is increasingly recognized that incidental findings are incompletely understood, expensive, and surprisingly harmful. Rather than a benefit of imaging, they are usually a harm. They are not sought, the odds of them being important is low, and they create tremendous uncertainty and low-value care. The pragmatic challenge is what to do about it in the near- and medium-term.
Some have wondered if certain incidental findings should not be reported at all. The medicolegal environment complicates matters. Some incidental findings are cancer. Sophisticated understanding of the biases which predict low-value care—that early detection of some cancers can produce a paradoxically worse outcome than had those cancers never been detected at all—is not a reasonable thing to expect of patients or the legal system in 2022; it is difficult for medical practitioners to understand. But we shouldn’t simply maintain the status quo. Here are several recommendations.
First, we should heed the call to action raised by some asking us to be more aware of the harms of overdiagnosis and overtreatment cascading from the detection of incidental findings. Incidental findings are a complication of diagnostic imaging—inadvertent harm despite positive intent—like bleeding following an image-guided biopsy. The specific harms of incidental finding management are opaquer than bleeding and harder to understand. But this simply means we (radiologists) should take a more active role in studying and managing them. It is our complication and our challenge to solve.
Second, we should advocate for incidental findings guidelines, especially our own but others as well, to explicitly incorporate and recommend appropriate studies to confirm they are working as intended. Working as intended means “producing high-value care”.
We should expect incidental findings guidelines to emphasize the creation of high-value care rather than an exclusive or overweighted focus on maximizing diagnostic sensitivity. This is not a radiology-only dilemma. Incidental findings guidelines exist in many medical and surgical specialties, and we should work collaboratively with them to promote a high-value approach.
Third, we should advocate for funding organizations to prioritize the study of incidental finding management. We have a compelling argument. Incidental findings are ubiquitous and an enormous burden on the health care system. Randomized trials could be conducted in which deferral of aggressive diagnosis and management is a treatment arm. The emergence of active surveillance as a valid strategy for many cancer types is a precedent we can follow, apply, and expand upon here.
Fourth, we should avoid being alarmist in our reporting. Yes, at present, we should follow the guidelines we support until stronger evidence arises, but we also should recognize that most incidental findings are not harmful if left alone in low-risk patients. Low prevalence of disease and the biases inherent to screening help explain why this is so. When in doubt about the significance of an incidental finding, and the guidelines are unclear or give leeway, err on the side of minimizing it.
Fifth, because the clinical importance of an incidental finding is highly dependent on patient risk, we should pursue information technology solutions, in collaboration with referring providers, to make relevant risk factors more visible to radiologists (e.g., hypertension uncontrolled on multiple medications [adrenal nodule], unreported head and neck cancer [liver lesion]). In current state, radiologists often rely on a brief historical snippet focused on the chief concern to interpret an examination. Incidental findings are definitionally unrelated to the chief concern and therefore not always informed by it.
Sixth, in our reporting, we should attempt to balance diagnostic sensitivity with other competing risks.
We should understand the cascading harm that can result from management of an incidental finding and allow that potential for harm to influence our recommendations. We are still largely ignorant about which incidental findings are important and how best to manage (or ignore) them. In the years between now and a clear solution, we should do our best to minimize collateral harm to the patients we are trying to help.