Diagnosis of cancer. Some nuances

Ilya Yasny, XXII CENTURY

Cancer diagnosis is an extremely popular and at the same time complex area of medicine. There are a number of misconceptions around it, and in this article I will try to dispel some of them, as well as show where this complexity comes from, what the current state of things and prospects are.

I hope that the article will be useful to many non-specialists, since almost everyone is faced with oncological diseases in life, and it is very important to understand whether those who are trying to deceive you by offering "cancer diagnosis by saliva" or "express diagnosis" are profiting from your fear and ignorance.

It is necessary to distinguish between screening and cancer diagnosis. Screening is a check of the general population, for example, during a medical examination or an independent checkup (a procedure when you are examined by several doctors). Diagnostic procedures carried out during screening can detect cancer or suspected cancer, and then the person is sent for diagnosis. Diagnosis, therefore, is the diagnosis of a specific disease in a narrower segment of the population with a high risk group.

Strictly speaking, all further arguments relate to screening, which is "popularly" often incorrectly called diagnostics.

Early detection of cancer is an extremely important task, perhaps the most important in oncology. Early detection of a variety of diseases, such as breast, prostate, colon, and cervical cancer, can cure most patients or at least bring the disease under control. True, with some malignant diseases, the situation is not so rosy, for example, the detection of pancreatic cancer even at stage I provides an average of 12-14% of five-year survival, but it is still better than 1% at stage IV.

However, the development of an effective screening test is not an easy task, as evidenced by a small number of such (despite the fact that there are more than a hundred types of oncological diseases). An effective test is a procedure, the implementation of which will significantly increase the detection of cases at early stages, start treating them in a timely manner and ultimately reduce mortality. For the general population, this is extremely difficult to prove, and then I will illustrate why. To date, this has been proven for five to six tests: X-ray for lung cancer, mammography for breast cancer (with reservations), gastroscopy for stomach cancer, colonoscopy and a blood test in the stool for colon cancer, Pap test for cervical cancer, visual examination for skin cancer. Tests for PSA and CA-125 proteins to determine prostate and ovarian cancer, respectively, remain a big question.

Why does even such an obvious method as mammography (breast X-ray) cause so many problems and uncertainty about the appropriateness of its use for screening? The thing is that any analytical method is characterized by parameters such as sensitivity and specificity, and they are never absolute, that is, they do not reach 100%. Sensitivity is the ability of the method to give a positive result, despite the fact that there is actually a disease. Specificity is the ability to give a negative result if there really is no disease. (In passing, I note that, thus, the determination of the sensitivity and specificity of the method relies on independent methods of detecting the disease, which themselves are flawed). Sensitivity and specificity are not independent values: as soon as we begin to increase the sensitivity of the method, its specificity decreases, and vice versa. Therefore, it is necessary to find a reasonable compromise between them.

Fig. 1. The horizontal axis shows the value given by the test result, for example, the concentration of the tumor marker protein. Vertically – the number of patients with this value. For any test, there is a so-called "gray zone" where the results overlap in healthy (white squares) and in sick (red squares). The parameters of the test – sensitivity and specificity - depend on the choice of the cut off value.

The problem arises when we try to use these imperfect methods to study large samples of patients in whom the prevalence of the disease is not very high. As a result, it turns out that even if the sensitivity of the method is 80%, and the specificity is 90.4% (which is very good, and these are the parameters of mammography), then out of 10,000 women tested during screening (assuming that 1% of those examined are really ill), the error of the method will lead to the fact that 950 healthy women the method will give a positive result. In addition to stress for the patient, this means additional time and effort for subsequent diagnosis. At the same time, 80 people will be found really sick, and 20 patients will be considered healthy (which is even worse). Thus, a positive result in the case of mammography only in 7.8% of cases (!) means cancer. Those who want to understand in more detail why this is happening, and what does Bayes' theorem have to do with it, I refer to two good analyses here and here.

In addition to the above problems, it should be borne in mind that fluoroscopy itself is unsafe and can provoke the development of tumors. Therefore, qualitative research is needed to show that the benefits outweigh the risks. Preventive studies to reduce mortality always require a large number of patients and very strict rules for obtaining reliable data. In Siddhartha Mukherjee's beautiful book The Emperor of all Maladies describes the thorny path that mammography has taken since 1963, which included dozens of studies on hundreds of thousands of women. As a result, mammography is now recommended for women over 40-50 years old every one or two years, and the benefits from it are very modest.

The situation is no less difficult with other tests, for example, with lung cancer. Studies conducted on more than 50,000 people have shown little benefit from the use of low-dose computed tomography compared to conventional X-rays, while most of those who have lesions on CT and who are sent for biopsy, cancer is not confirmed.

Relatively good results against this background are provided by screening programs for cervical cancer and cancer of the rectum and colon (colorectal cancer). Colorectal screening is reliably effective, because when diagnosed at early stages, the 5-year survival rate increases from 68% to 90%. The population studies that showed such a benefit of screening included about 50,000 participants and took 15-20 years. Early detection of cervical cancer by smear analysis increases the 5-year survival rate from 30% to >90% and reduces the incidence of invasive cervical cancer by 2-6 times. This was established after analyzing registries around the world before and after the introduction of smear screening.

All other cancer screening methods, such as the determination of prostate cancer by PSA protein, ovarian cancer by CA-125, various options for testing proteins, microRNA, gene, transcription and epigenetic material of patients in various combinations are not yet suitable for screening the general population due to the problems described above with sensitivity and specificity of methods. However, some of them are used (and others are being developed) for use in specialized oncological dispensaries and clinics for diagnosis as such, that is, for making an accurate diagnosis, for prognostic purposes, for monitoring the success of therapy.

It should be emphasized once again that the benefits of the testing method cannot be considered proven if only its analytical parameters, such as specificity and sensitivity, are shown on samples taken from patients with a known diagnosis. Large, high-quality studies of the test in real conditions are needed, when patients are first tested with a new method, then the diagnosis is confirmed by the old one, and only the old method is used in the control group. Then they measure how different survival rates are in these two groups. Since the average survival rate can reach 10 years or more, and the analytical parameters of the methods inevitably deteriorate with the transition to larger cohorts of patients, studies require so many people and turn out to be so long.

Is there any hope for an improvement in the situation, and what developments are currently underway in this area? There are quite a lot of developments, but most of them are aimed at improving the analytical parameters of existing screening methods or at using them in patients with a positive screening result (with suspected cancer) in order to reduce false positive results. There are also completely new developments aimed at screening those types of cancer that are poorly detected. However, nothing is visible on the horizon of five years that would allow us to identify diseases such as kidney, pancreatic, liver, stomach, brain cancers in the early stages: these cancers are still detected randomly, and at the same time they are often aggressive. Earlier detection would significantly reduce mortality from them. Perhaps one of the promising methods of future screening and diagnosis of malignant diseases will be the study of exosomes – vesicles that split off from cells, including cancer cells, and include part of the contents of the cell. But it will take many more years before the research, which began recently, will answer the question of the usefulness of this and other approaches.

Fig. 2. Exosomes circulating in the blood are split off from cells and may contain proteins, RNA, DNA of the original cell. Studying their contents may make it possible to diagnose a number of diseases without the need to apply invasive procedures.

Unfortunately, the fear of oncological diseases, coupled with the lack of awareness of people and the unequal quality of oncological care (especially primary care), allows unscrupulous firms offering services for the "diagnosis" of cancer to multiply. Actually, I was prompted to write this article by a controversy with the developer of one of the screening tests – a saliva cancer test. Despite the fact that the developer has no publications in international journals, there are no research results on the actual implementation of the test, there is no permission to use the test as a diagnostic, he shamelessly sells such a test for several thousand rubles, and only in the contract of the public offer, which must be downloaded separately on the website, you can find a mention that that the test is purely informational in nature and is not intended to inform medical decisions. I would like to warn readers against falling into the hands of charlatans: there is no way to identify all malignant diseases at once, and even distinguish them from each other, and will not appear for a long time. There is also no "express diagnosis" of cancer, as there are no cancer tests based on a drop of blood, saliva, the line of the hand and the iris of the eye.

A set for taking a smear of one of the many methods of "diagnosing oncology by saliva".

Finally, what should an ordinary person do in this situation? It is necessary, firstly, to find a good therapist who has been recommended to you by trustworthy doctors. Secondly, to determine whether you belong to any risk group and whether you need to undergo regular screening. And only after that it makes sense to spend money, nerves and strength and begin to be systematically checked for malignant diseases.

Portal "Eternal youth" http://vechnayamolodost.ru  15.07.2016