Nuclear Medicine and Artificial Intelligence Combine Against Prostate Cancer | the daily

Prostate cancer is one of the most frequent in men in Uruguay. On average, there are 1,466 new cases per year, according to the latest data available from the National Cancer Registry (2013-2017), and it is the second in mortality –after lung cancer–, with an average of 573 annual deaths in that period . The sooner it is detected, the better the prognosis and in that, as well as in improving the treatments, professionals from the Uruguayan Center for Molecular Imaging (Cudim) work, including the nuclear doctor Gerardo dos Santos and the physicist Enrique Cuña, who received the daily at the center as part of Blue November, a global campaign to raise awareness about prostate cancer.

The Cudim was inaugurated in 2010 and is the only center in the country that performs positron emission tomography (PET) scans. Dos Santos synthesized that this study is a nuclear medicine tomography that, from the injection of radiotracers –which emit radiation, positrons–, allows us to observe the metabolic functioning of different organs. As the equipment that makes the PET allows, at the same time, to make a conventional computed tomography (hence it is called PET-CT), “a technique that is hybrid, a morphological and functional image” is obtained that improves the identification of anomalies .

PET diagnostics

A PET scan is not done in the first instance to detect prostate cancer. What is recommended is that men between 40 and 50 years of age know, through a blood test, the level of prostate specific antigen (PSA), which is a substance produced by the prostate; in case of doubt, this is complemented by digital rectal examination. Dos Santos added that if these results indicate that something is out of the ordinary, conventional imaging studies are usually indicated, such as a tomography, an MRI; a biopsy puncture completes confirming the diagnosis.

“The primary treatment in prostate cancer can be surgery or radiotherapy, which are supposedly curative treatments and have good efficacy, but unfailingly at a time of evolution a significant percentage of patients, up to 30%, may present a relapse of the disease. illness after being treated ”, explained the doctor. This relapse – it is called a “biochemical relapse” – is detected by a PSA blood test. To study it, he said that conventional imaging techniques can be used, such as tomography, bone scan or magnetic resonance imaging, but that “all of them separately sometimes have low sensitivity to be able to diagnose relapse and PET plays a fundamental value , because it can detect with greater sensitivity where the disease is ”.

How is the disease detected through PET? “In nuclear medicine, molecular targets are sought, that is, structures that in some cells are overexpressed in greater amounts in cancer cells than in normal cells. It has been shown that in prostate cancer there is an overexpression of a glycoprotein called PSMA, prostate-specific membrane antigen, ”explained Dos Santos. The radiopharmaceutical that Cudim uses to detect them is called 68Galio-PSMA, which emits positrons and targets the PSMA.

“To make the diagnosis, the patient is injected intravenously with the radiopharmaceutical, we wait an hour for it to be distributed and then the PET study is obtained, which allows us to know where the disease is,” Dos Santos explained. The doctor clarified that it can be detected inside or outside the prostate; in the latter case we speak of metastasis, because the tumor has spread to other areas.

Biochemical relapse is the fundamental requirement for someone to receive a PET scan, which re-stages the disease (again expresses the disease state). Dos Santos argued that PET restaging is covered by the National Resource Fund (FNR) and added that for some months the FNR has also covered the indication of PET for initial staging in patients with recently diagnosed medium-risk prostate cancer. high.

The exact location of the lesions allows for targeted radiotherapy or local surgery on the affected area, such as a bone, and improves the survival of the patient.

PET treatment

In addition to making diagnoses, the Cudim does treatments and research. In prostate cancer, the center is doing treatments with a radiopharmaceutical called Lutecio 177. “Unlike gallium, which is a positron emitter, lutetium emits much more powerful energy and allows us to specifically treat tumor cells, as it enters the malignant cell, while healthy cells are protected “Dos Santos pointed out, adding that” it is an effective treatment in patients with many metastases and who do not respond to classic treatments, such as hormone therapy or chemotherapy.

Dos Santos clarified that “to do treatment with lutetium we first have to carry out a PET study with 68Galio-PSMA”, which is what confirms that the lesions capture the radiopharmaceutical. Once this is verified, lutetium is injected instead of gallium. This novel technique is called “diagnostotherapy” or “theragnosis”, which is “the union between diagnosis and therapy”, he said, adding that a “targeted metabolic therapy” is done with which, unlike chemotherapy, you know exactly which sites you are dealing with.

Dos Santos explained that in Uruguay lutetium treatment began to be used in 2017 in a clinical research project in which several institutions participated. He explained that the patients were recruited at the Hospital de Clínicas (where Dos Santos is a teacher), they were controlled at the National Cancer Institute (of the State Health Services Administration) and teachers from the Faculty of Chemistry participated. do the dosimetric calculations – and the Cudim chemists, who did the quality controls when the lutetium arrived from abroad.

This project involved 17 patients and ended in December 2020. The treatment “consists of at least three intravenous injections that have to be six weeks apart from each other; After the second, a PET study is done to corroborate a response, which is generally favorable, and there the third dose is applied. If there is a very good evolution, additional doses can be made, up to five or six is ​​a safe range ”, the doctor explained.

Of the 17 patients, the majority received three doses, but in some cases up to five. “The treatment is very well tolerated by most patients,” Dos Santos said, specifying that adverse effects are rare and mild, such as tiredness or a feeling of dry mouth. A slight decrease in white, red or platelet blood cells is also expected, which is controlled with blood counts. The person receiving the treatment must have good kidney or liver function values.

Dos Santos highlighted that more than half of the patients reported a significant improvement in pain, which occurs when there are bone metastases – something that also limits mobility. A decrease in PSA and the number of PET lesions was also observed. The doctor appreciated that all this determined an improvement in the quality of life of the patients.

These 17 people received the treatment at no cost, because they were part of the research. The patients who receive it now, from Uruguay and abroad, pay for it privately or privately, Dos Santos said. Each dose costs about $ 7,500; Cudim only charges lutetium, according to the researchers, the service is non-profit and that is why it is “much cheaper than in other countries.”

“This puts Uruguay at the forefront, because they are treatments that did not exist before and that were made here for the first time. We are on par with the first world, both in treatments and in diagnoses, ”said Dos Santos.

Artificial intelligence

From his area, medical physics, Cuña explained that “PET technology has to do with purely physical principles” that include “the emission of radiation caused by the radioactive substance when it is deposited in metabolic centers, in catchment centers” , and the radiation detection that PET cameras do. “The images have much more information than what can be captured with the naked eye,” said Cuña, and for that he appeals to artificial intelligence, specifically, to a field of medical study “that is in vogue” called radiomics and seeks to extract information from images through artificial intelligence tools. Cuña explained that the heterogeneity of tumors can be seen from the study of the images and the construction of “a map of the textural characteristics” from which “texture indices” can be devised to be detected by an algorithm. computer scientist.

Cudim has not yet applied the tools to prostate cancer imaging, but it is on the way. Until now, Cuña has worked with a database of foreign head and neck cancer patients and with another Cudim database of pediatric patients with lymphoma.

“In the database of patients with head and neck cancer, the textural characteristics in the tumor areas were measured and then the correlation was made; we have the images of the patient before and after the treatment and with these indices, with these textural characteristics a relationship was sought that could indicate a response to the treatment, ”he explained. The differences are seen through the gray levels of the image pixels and a classification is made from “thousands of data”. “That mass of data is passed to an algorithm, and the algorithm by artificial intelligence methods – what is called machine learning– is able to classify the patients. That is called a machine learning supervised, that is, I already know, because I have the database, which patients responded and which did not, “said Cuña. It is about “training the algorithm”: it passes it the data of a group of patients and asks it to develop a model that predicts which group each will be in. “Once the algorithm is trained, the group of patients that was left out is passed to it and it is tested to see how well they respond,” he added.

With this study, Cuña found two indices of textural characteristics that may indicate a response to treatment. He remarked that they are indices that “are difficult to visually interpret” by humans: the machine will do it better, he acknowledged.

The database of lymphoma patients was small, 21 people, but it had the advantage that the Cudim had followed them extensively, of around four years. The researchers used the same textural characteristics from the other database and sought to predict the response of lymphoma treatments; they passed that information on to the algorithm and found two textural characteristics “that could potentially predict a kind of response to treatment.”

They will seek to apply this same work scheme in patients with prostate cancer. “Work is being done in collaboration with a Spanish company called Quibim, which has an online platform prepared to upload images and do analysis with textural characteristics,” Cuña reported, adding that they are using magnetic resonance images, which have higher resolution than PET. “With this high resolution, they have developed algorithms to measure these biomarkers and also to segment the regions of the prostate; For this, something called ‘neural networks’ is used, which is part of a field of study called deep learning, which is a neural network, with decision nodes that interconnect with each other, and it is a mathematical model that I can train to make certain decisions ”, stated Cuña; The images are uploaded to the platform for the algorithm to segment the areas of the prostate that they use to stage cancer in a classification system – called PI-RADS – that influences treatment. “It is a support tool for doctors and radiologists, it is something completely computerized that can be used with medical images to streamline work and. in the case of these textural characteristics or biomarkers. they can be a complementary tool for a possible decision about the course of treatment that can be recommended to a patient ”, he concluded.

Source link

About Admin

Check Also

Inauguration of the Resilio Behavioral Medicine psychiatric center – Diario Social RD

With the assistance of medical professionals, medical representatives, collaborators and friends, the Resilio Behavioral Medicine ... Read more

Leave a Reply

Your email address will not be published.