Would be Point of Care Testing or Laboratory Medicine

1. Dr. Oliveira, the field of pathology has evolved significantly over the years. Could you outline the major milestones in the evolution of pathology, particularly in the integration of morphology, immunohistochemistry, and genetic sequencing?

Pathology has experienced a major evolution in the last two/three decades. Just simple morphology observed by H&E staining is insufficient since we need to extract more information from tissue to provide definitive diagnosis and targeted therapies. In the last years, morphology has regained some extra attention, since “old details” that were perceived, but not mentioned have been giving prognosis value, such as percentages of specific tumor components (Gleason prostate grading), tumor extension, morphological patterns (MELF pattern in endometrial cancer, Silva pattern in cervical cancer, histological growth patterns in liver metastases) and as such, they need to be mentioned in the reports.

Implementing routine immunohistochemistry for diagnosis in the last years of the XX century was a milestone since it allowed the correct classification of tumors, based on their immunohistochemistry profile, prompting a detailed classification and even suggesting biomarkers for therapy, such as HER2. Nowadays almost every oncological specimen is studied resorting to one or more biomarkers, for classification (ex: use of TTF1 and P40 in lung cancer) and treatment purposes (ex: HER2, IMS testing, PD-L1). The integration of genetic findings was also a breakthrough. Firstly, by identifying point mutations by simple PCR and nowadays with larger sequencing panels, it is already an integrated part of routine pathology. The need to know more and in detail leads us to the use of more sophisticated tools, which “question” our biological samples and provide us with more complete answers. This provided insight into mutations and gene fusions, allowing further characterisation and classifying of the diseases. It is a daily routine component of a soft tissue pathologist since a great part of the entities are defined by gene driver-specific mutations, and are no longer possible to assess by simple morphology and immunohistochemistry. The genetic classification has also implications in prognosis, and selecting the correct therapy, and there are cases where the presence or absence of mutations implies targeted therapy – EGRF mutations, KRAS mutations, NTRK, and FGFR2, among others. The creation of Molecular Tumor Boards worldwide is an example of genetic importance.

Therefore, a modern pathologist has to integrate multidisciplinary knowledge, based on a first observation – H&E or special stains morphology, complemented by immunohistochemistry and refined by molecular pathology. Only this full integration would pave the road for individualized and precision medicine.

2. How do you see the role of pathology today in providing a more precise and personalized approach to diagnosis and targeted therapy?

Pathology is fundamental. The biological sample analysis is the cornerstone of diagnosis. Despite the multiplicity of non-invasive methods available, tissue sampling is fundamental for treatment decisions and therapy institutions. Several inflammatory/autoimmune diseases may present a clinical conundrum or be mistaken as oncological diseases, such as an IgG4 biliary stenosis (radiological interpreted as a cholangiocarcinoma), and only tissue sampling will be able to define the correct aetiology, with different treatment options. A tissue examination is also necessary for defining the correct treatment. Nowadays every lung cancer is tested for PD-L1 expression, and every breast cancer is tested for hormonal status and HER2; molecular analysis is the actual basis for the Central Nervous System tumors classification, stratifying treatment and prognosis. Therefore, “tissue is the issue” and puts pathology at the center of precise and individualized medicine.

3. In your view, how has the growing demand for faster results influenced the pathology field, and how have laboratories adapted to meet these demands?

There is a growing demand for faster pathology results; however, there is also a lack of pathologists worldwide. This was analyzed and published in the journal The Pathologist, by Andrey Bychkov and Michael Schubert, in a paper called “Constant Demand, Patchy Supply”. Pathological examination, as we saw in the previous answers is becoming more and more complex, therefore requiring more time for analysis. The need for immunohistochemistry and molecular analysis also adds extra time to the report TAT, but as explained this is a fundamental process.

Laboratories have tried to adapt to new bigger and faster equipment, by implementing new and faster protocols, Lean management and development of more functional laboratory workflows, and integrating all of the diagnostic procedures. Some historical manual tasks, such as embedding and slide cutting, have been automated (but still without the speed/profitability of manual proceeding), but there is still a long way to go. The possibility of longer work shifts is a major value in some developed countries, with 16 to 24h daily work, allowing for faster TAT, but there is still a high need for professionals in pathology, which makes it difficult for universal implementation of these measures. In oncology, we are always working against the clock, but one must stress that although a fast report is important, a correct and detailed report is more important.
In our lab we optimized the flow due to digital pathology implementation and it needed some major changes. It is always a hard work and needs commitment from the entire staff.

4. Point of Care Testing (POCT) is increasingly popular due to its ability to deliver quick results. Do you think POCT will eventually replace traditional laboratory testing, or do you foresee it as a complementary tool?

In pathology, it is more of a complementary tool. The pathological process involves several steps such as tissue fixation, tissue embedding and cutting, staining and observation. This makes it difficult to implement a POCT. There are cases when we can do short protocols for quick answers, such as kidney transplant evaluation and heart rejection grading, but they are limited. Frozen sections are also fast, but they are usually employed for intra-operatory procedures. Probably the best situation that we can use in pathology and is like POCT, would be a Rapid On-site Evaluation (ROSE) procedure. It is not a POCT since it does not provide a result, but it allows for in-site evaluation of the sampling, assessing for its content and classifying it as adequate or unsatisfactory. In the latter situation, one may do extra sampling, therefore reducing the number of unsatisfactory results and consequently reduction of new and redundant appointments. ROSE was devised mainly for cytology, but similar procedures may be employed in tissue biopsy, assuring that the tissue amount is representative and enough for the procedures.

5. What are some of the challenges involved in integrating advanced techniques, like genetic sequencing and immunohistochemistry, into the routine diagnostic workflow, especially in a high-volume setting?

There are several challenges in an integrative diagnostic workflow. With a high volume, the demand will increase in all the sectors. The main need would be physical space and equipment. Pathology laboratories need a large space for their routine settings. Even in digital pathology laboratories, the laboratory's procedures need a proper physical space, enough for machine accommodation and safety measures (formalin extraction, among others). The need for a dedicated and experienced team is also a requirement, since with a high volume the laboratory flow has to go smoothly and with the less amount of interruptions. Experienced staff is necessary since it allows faster and more accurate tasks, but it is also a challenge due to the lack of pathology professionals. The laboratory flow needs to be finely tuned. As stated before, a precise report (especially in oncology) is expected to have immunohistochemistry testing (one or two rounds of testing), so the time between observation and testing has to be as short as possible. This implies enough machinery; otherwise, one must wait for an empty slot for testing purposes, which would affect the TAT. This is also applicable to genetic testing. Genetic testing takes some time – from 3-5 to 20-30 working days, depending on what we are looking for. The machines have a limited amount of space and once they are running we must wait for the end of the cycle. This stresses the need for an adequate number of machinery and staff and to have a finely tuned and sequential workflow. Some countries have tried to solve this issue by creating reference centers, which worked in the beginning, but with the rapid increase in the workload, they became overloaded very shortly.

6. Given the advancements in artificial intelligence (AI) in medical diagnostics, how do you envision its role in pathology? Do you think AI will significantly impact diagnostic accuracy and efficiency?

AI is everywhere and pathology is not an exception. With the process of digitation, the data are waiting to be mined and analyzed. Several systems for AI are being developed and implemented and after the Paige approval by the FDA, others will shortly follow. The current AI applications would be of high value, since they have the potential of organizing our work, prioritizing samples with a high probability of malignant disease, pre-analyzing samples and providing hot spots for counting, etc. Diagnostic accuracy can be improved using this software, but it should be just a minor increase since a trained pathologist should have good diagnostic accuracy. The main value here would be assessing time-consuming tasks, such as counting mitosis, biomarker quantification, improving the TAT and efficiency and, rendering reproducibility, which is a main concern in pathology. Several analyses are based on hot spot evaluation, which can be selected differently according to the pathologist. The identification of such areas with precision by AI would render the same results everywhere.

7. How can AI algorithms assist pathologists in dealing with large datasets, especially in the context of genetic sequencing and complex immunohistochemical analyses?

AI in large datasets would be highly valuable since it would have the capacity to generate data otherwise very difficult and complex to obtain. Nowadays tumor microenvironment analysis is a hot topic, and several mechanisms are available, such as multiplex immunohistochemistry or fluorescence. This is a process where in a single slide, 8 or more makers are available at once. There is the need to transform all that information into a value or a classification prompting clinical clustering and decision making. Pathologists are used to work with immunohistochemistry, but we work with one (maybe two) biomarker in each slide. Putting all the information in one slide would be mind-blowing. AI would certainly have a major role here.

Genetic sequencing is becoming more and more a reality, and with the exception of single gene testing for classification purposes, the majority of oncological samples are going to complex genetic panels in order to find actionable mutations for treatment. The interpretation of a genetic report is not always easy, in a field where information changes every day. AI should be able to provide quick answers by analyzing data and finding clinical trials or approved drugs, allowing the information to be discussed at dedicated boards.

8. What impact do you believe AI will have on the future of laboratory processes, particularly in automating repetitive tasks and improving turnaround time?

AI would be extremely valuable in improving TAT. Straight the pathologist by aiding in time-consuming tasks, such as cell counting, mitosis identification and biomarker quantification, would allow precious time savings. AI in report construction is also to be considered, with automatic template production and automation filling data such as patient ID, clinical information, and attending physician ID, among others for the pathology order document. In the lab, things are not so easy, since it is a manual-dependent setting, but several solutions are being developed for embedding and cutting. Grossing is also time-consuming since a lot of measurements have to be performed; some systems automatically weigh and measure the samples allowing some time reduction. One must not forget the role of AI in analyzing the LIS data. AI has the potential to scope the data from the LIS, assessing for time consumption in each task and suggesting ways to save time by reorganizing lab workflows.

9. There are concerns about the potential for AI to replace pathologists. What are your thoughts on this, and how do you think AI can be integrated in a way that complements, rather than replaces, the role of the pathologist?

I foresee a limited implementation in the near future. Due to IVD (in vitro diagnosis) regulations, all AI systems must follow this rule. In addition, since there is no universal AI algorithm, the labs would need an AI for the breast, an AI for the lung, an AI for the colon, etc. This would require a major amount of software just for a routine basis, which would be highly expensive and difficult to refund. There are also ethical and legal questions involved in the total replacement of a pathologist in rendering a diagnosis.

I believe that AI has a major role in complementing the pathologist by aiding it in focusing on suspicious areas, solving time-consuming processes and improving the reproducibility report, more than replacing it. A pathologist working with AI would be a more valuable asset than a pathologist that do not use it, but there should be no replacement. The time-saving for the pathologist would provide more report signing, research and more availability for therapeutic decision meetings.

10. With the growing importance of molecular diagnostics, how do you see the future of pathology departments in terms of integrating new technologies and staying ahead in the field?

Molecular pathology is a component of routine diagnostic as is immunohistochemistry. It is no longer possible to have a good pathology report without molecular analyses for diagnostic purposes (in sarcomas for example) and for treatment decisions. Especially in the treatment modalities, this is necessary. Every day new drugs are emerging and we need to keep it up to provide our patients with updated and individualized medicine. More sensitive tests are required, with wider genetic coverage and, especially, faster, since in the oncology field we are always running against time. New technologies must be integrated; otherwise, a pathology lab would become obsolete.

11. How do you think the current regulatory environment is adapting to the rapid pace of innovation in the pathology field? Are there sufficient measures in place to ensure accuracy and reliability?

Despite the rapid pace, I believe that there are measures that assure accuracy and reliability. The devices have to be approved by local and international agencies that have high standards and when introduced in the lab they undergo internal validation protocols. Additionally, the labs are expected to keep up with External Quality Assessment (EQA) on an annual basis, ensuring the correct functionality of their equipment. Only with strict quality, and performance control is possible to perform quality and individualized medicine. We must take into consideration that it is based on our results that patients are getting their treatments. We must be sure of the quality and integrity of the results. It is a fundamental process and quality should be always in the laboratory standards.

12. In clinical trials, pathology plays a key role in assessing treatment efficacy. How has the integration of advanced diagnostic tools impacted clinical trial designs, particularly in oncology and personalized medicine?

Clinical trials are fundamental in medicine and are becoming more complex due to increasing pathophysiology knowledge. Nowadays clinical trials require more data, but also more genetic material since patients need specific biomarker analysis – immunohistochemistry or molecular. This has driven the clinical trials to be centred in highly advanced clinical centers, since they are the ones that (usually) possess the technology (and knowledge and data interpretation) for this kind of analysis. Therefore, clinical trials are more and more centered in advanced tertiary centers, which is good for reducibility, but also limits the chances of minor centers being a part of it. Nevertheless, digitation and AI analysis can be a valuable tool in harmonizing patient samples for clinical trials.

13. The demand for real-time diagnostic data is increasing. What measures can be taken to ensure that laboratories maintain the highest standards of quality while meeting these demands?

Once again, we are heading to the pressure of having the fastest results for the patients. Special caution has to be taken when we fasten our results since an assertive report is more important than the fastest one. Laboratories must ensure their quality by checking their workflow and participating in EQAs. Having external certifications and accreditations is also a good method for quality checking and improvement; however, this process takes time and has costs. Laboratorie boards must be aware that spending time and money on quality is not spending at all, but winning in quality and one way to ensure patients have the best possible results.

AI should be a valuable partner in analyzing and interpreting real-time diagnostics, but this AI process must also be analyzed with quality. The concept of explainable IA is gaining momentum and should be a regular partner in laboratory’s daily routine.

14. Finally, how do you see the future of pathology and laboratory medicine evolving in the next 10-15 years? What innovations or trends are you most excited about, and what challenges do you anticipate?

We will probably go deeper in analysis and gain more insight. Individualized medicine is our goal and therefore we must evolve more and combine more technology. Liquid biopsies should be a daily test, especially in early cancer detection and disease monitoring such as minimal residual disease (MDR) detection of point key mutations for treatment shifts, with increased sensibility and sensitivity. Different genetic tests, encompassing more and more actionable genes would be available and with faster TAT, preferably with a POCT methodology. The integration of different AI solutions will augment the pathologist's capacity to render more precise diagnoses. Fully automated laboratories using robotics will result in 24-hour open labs and reduced errors. With increasing disease pathophysiology, the integration of multi-omics and big data will be on a daily basis, providing a holistic view of diseases with genomics, proteomics, lipidomis, radiomics, etc.

The main challenge will be keeping up with innovations. This methodology would demand labs to grow, physically and in terms of human resources, and as we have discussed there is a shortage of professions in this field. Physically labs would need to accommodate more and more machinery or establish network partnerships allowing a complete workflow. Finally, the pathologist as we know will have to differentiate into several subtypes in order to provide updated knowledge and real-time decisions, providing a more deep subspecialization: there will be the interventional pathologist (a more or less fusion of interventional radiologist with pathologist would be able to take tissue and sampling on the spot ensuring enough materials is obtained), the molecular pathologist, the genetic pathologist, the AI pathologist, the bioinformatic pathologist, among others. This subtyping may have a reflection inside the lab, providing the need for more specialized laboratory staff.

--Issue 05--