TrailblAZers in Oncology:
Transforming patient outcomes through computational pathology

I loved Saturday specials on the BBC World Service as a kid. I never could have imagined this childhood pastime would set me on the path to discovering my passion for patient care and life’s work. It started with a special on the US Centers for Disease Control and Prevention (CDC). I had never heard of the CDC and was mesmerised by the way the programme described chasing down patient zero at the heart of a pandemic and seeking ways to ease human suffering through medicine.

I was a lonely, awkward kid who loved to read. I read constantly with no predilection for any particular genre, even reading the biology textbooks of my older siblings. A large part of my reading included mysteries, anything from Agatha Christie to Sherlock Holmes to the Nancy Drew and Hardy Boys series. Mesmerised by the BBC special, a lightbulb went off for me: “They are the disease detectives!” And right then, I knew I was going to be one too.

Pathway to digital pathology

I grew up in India and saw the need for disease detectives all around me. Devastating diseases like leprosy and tuberculosis were endemic where I grew up; it was not uncommon to see lepers amongst the pedestrian traffic, and to have the word ‘tuberculosis sanatorium’ in a 6-year-old’s vocabulary is testament to the prevalence of this disease. The word cholera caused grave anxiety for my mother. I later learned she lost her favourite sister within a matter of hours to a cholera outbreak and could not say goodbye due to the quarantine measures. Every day I was acutely aware of the suffering these diseases caused. I went on to study microbiology and received my master’s degree with distinction and honours. However, I was frustrated at the lack of wet lab experience that would have allowed me to put my studies to use. In addition, I was painfully aware an isolated upbringing with my nose often in a book did not exactly prepare me for the world. I decided to put on hold any plans for further education and said yes when I was offered a position as a junior research fellow at the National Centre for Biological Sciences (NCBS). I was a misfit for lack of experience but approached this opportunity with grit, determination and persistence. It was not easy, but I understood what could happen if I applied my drive and ambition with a commitment to the long hours. I worked for four years in a community of researchers that asked the simple question why and how do cells die? Those formative years taught me to listen, to verbalise my curiosity, to design experiments, interpret the results and communicate with my peers, as well as to decide and prioritise what to follow on. There was hardly a day that went by without a marvel on the bench proving a hypothesis whilst opening a door to an idea or an intrigue. Never had I experienced the very unique euphoria that comes from connecting the data points and completing the story with milestone experiments standing out like pearls on a chain.

When I did eventually receive my PhD, I ended up at a crossroads. I did not want to follow the same path as so many others before me — I craved a change that would lead me down new paths of exploration. So, I took another chance and joined a digital imaging and pathology department at a pharmaceutical company that would transform my life in ways I could have never imagined.

Digital pathology: a whole new world

I had next to no experience in pathology, let alone digital pathology, but that was not a deterrence. The four years I spent with this group were similar to the formative four years I spent at NCBS. I was a sponge, absorbing knowledge of the pathology lab, digital pathology techniques, learning about the health platforms and the quantification of drug responses through image analysis in patients. The evolution of technology and its applications were absolutely astonishing, and so was the realisation of what we were actually accomplishing.

Pathology reignited my childhood passion, giving me new insights into the impact of scientific research and how mysteries of diseases are solved. The potential for patients was truly inspiring, but there was one big frustration for me: how pathology is practiced has remained largely unchanged since the 19th century.

Almost all cancer diagnoses are made or confirmed by pathologists through learned interpretation of morphological changes observed via an optical microscope, mostly from H&E-stained tumour samples mounted on glass slides. This 19th century method, which relies on subjective pattern recognition, remains the gold standard. Pathologists are expected to also provide accurate quantitative measures such as tumour grade, proliferative indices and intensity of antibody staining, which are prone to subjective error. Increasingly, such expectations require precise measurement of complex heterogeneous and diverse patterns that are beyond the capacity of the human brain to analyse accurately. While these measures are not always consistent, they are heavily relied upon for assessing the best potential treatment for people with cancer. The digital transformation occurred in radiology over 15 years ago, but a sufficiently compelling business case proved elusive in pathology because the gold standard method is highly cost-effective. Artificial intelligence (AI) can help transform these practices and will not only change the practice of pathology but will create massive imaging databases that can be mined to identify previously-unrecognised patterns that will provide new biological insights, improve diagnostic accuracy and accelerate the adoption of precision medicine.

Technology was rapidly advancing, and I knew this is where I could make my mark, helping to accelerate the process of getting new treatments to patients in need. Digital and computational pathology was the answer.

Transforming outcomes with digital and computational pathology

Digital and computational pathology have the potential to transform the field of medicine in unprecedented ways – by changing the practice of clinical drug development and driving the development of the next generation of precision medicine diagnostics.

A key part of putting patients first is choosing the right drug for the right patient at the right time. My team uses AI-driven methodology to extract objective information from tissue, utilise this information to predict responses to treatment and provide clinically-actionable conclusions to improve patient selection and develop personalised treatments. Our goal is to build more powerful, quantitative tools that equip drug development researchers with more clinically-actionable information than we could gain with the human eye. We have the ability to quantify millions of cells and the sum total of all their spatial relationships in exquisite detail in minutes. And that is just the beginning.

The power of computational pathology approaches lies in its ability to provide objective, precise, and robust assessment of a tissue biomarker expression and localisation. In addition, they are broadly applicable, combinatorial, scalable, faster and deliver a more accurate diagnostic with increased reproducibility and greater predictivity. With technology evolving every day, the algorithms we are developing are becoming even more powerful and bridging the gap to achieving our goal for the patient.

One computational pathology approach is our flagship fully supervised algorithm: Quantitative Continuous Scoring (QCS). QCS not only looks for the presence or absence of a biomarker — an important indicator for targeted cancer treatments – but it identifies all tumour cells and quantifies the target biomarker at the pixel level in the entire sample. This has the potential to be the first AI-driven test, providing countless benefits to the patient. By making the primary diagnosis process easier and faster, we can ensure patients are receiving the right treatment for their specific cancer quicker than ever before.

Enhancing the human element behind the data and computational science

I have been able to work on these incredible projects in my current role because once again, there was the desire for a challenge and the drive to better myself in that journey. I was hired by a small biotech company that was later acquired by AstraZeneca and it was once again a transformative moment of enormous proportions for me. In addition to witnessing the growth in the field of digital and computational pathology, I have experienced my own intense growth. I went from the role of a senior scientist to a senior director in less than a year and acquired new skills and expertise that reminded me of the important role people play in medical research and discovery. Technology is a vector through which we enhance our abilities to identify, treat and cure diseases, but the driving force behind discovery is people. I am fortunate to have worked with such a talented and dedicated pool of skilled individuals, many of whom were my best resources for learning and shaped who I am today.

People continue to be responsible for all we accomplish in this field. There may be patterns beyond the capacity of machine learning and the human brain to compute, yet for all of its amazing qualities, AI still relies on input and interpretation from human minds. There are many questions still to be answered about AI and its role in our world going forward. When it comes to computational pathology, however, I see a future where AI transforms pathologists into computational pathologists.

Shaping the future of digital and computational pathology

It has been a long time since I have read mystery fiction, but often times ponder about how those stories connect to the mysteries we continue to solve in research. Digital and computational pathology enables researchers to take on a role as a “disease detective,” which is more important than ever, and we now have so many new tools, computational methods and technologies to superpower our work.

It is a privilege to build the algorithms and steer the development of sophisticated software and tools that will change how computational pathology will improve patient lives. We are on the very edge of a bright new future for patients, and I am so proud to be part of the team at AstraZeneca that is leading the way.

Hadassah’s story is part of our ongoing series: TrailblAZers in Oncology, which spotlights colleagues on the AstraZeneca Oncology team who are making an impact, both personally and professionally.

Veeva ID: Z4-63221
Date of preparation: April 2024