05 March 2019
Share this article

Finding new answers in old medical records

Do all young women diagnosed with breast cancer really need adjuvant cancer treatment such as chemotherapy or hormone therapy after surgery to lower the risk of recurrence? Or would some be better off without it?

To find the answer to this question, Gwen Dackus (NKI/AVL) digs into the historical patient files and biobanked tissue samples of young breast cancer patients who didn’t get adjuvant therapy. Working with medical records stored in the IKNL Netherlands Cancer Registry (NCR), pathology data and tissue samples accessed via the Netherlands PALGA database, and advanced research tools provided by TraIT – one of the founders of Health-RI – she hopes to be able to give better advice to young women diagnosed with breast cancer. 

Image to the left: Gwen Dackus (NKI/AVL)

Before the year 2000, few breast cancer patients under the age of 40 who were diagnosed solely with a primary tumor received adjuvant chemotherapy or hormone therapy. Physicians believed that these adjuvant therapies were only necessary for women with lymph node metastases. Nowadays, doctors recommend adjuvant therapy for most young women diagnosed with breast cancer, because breast cancer tumors in younger women tend to be more aggressive. The progressive use of adjuvant therapy since 2000 has a resulted in an increase in the percentage of young women who survive breast cancer. However, the increase is only slight. This raises the question Gwen Dackus and her team at the National Cancer Institute are trying to answer: Are some women getting unnecessary treatment?

“With our research we want to gain knowledge that enables physicians to predict which breast cancer patients have a good prognosis, because those patients might not need additional treatment,” says Dackus. “It’s important because these treatments have severe side effects, which you’d rather avoid if possible.” 

With almost two decades having elapsed since the last group of young breast cancer patients were treated with surgery only, researchers have the opportunity to examine the differences between those with a good long-term outcome and those with a poor outcome. Working with the Netherlands National Cancer Registry, Dackus and her team tracked down 3,500 young female patients diagnosed with breast cancer between 1989 and 2000, together with their disease characteristics, treatment regime and long-term results. The accumulated data was analyzed to answer questions such as: did their cancer came back within ten years, and if not, did it mean that their breast cancer never returned? 

Valuable research facilitated by good record keeping 

The Netherlands is one of the best places to do this kind of research, because everyone diagnosed with cancer is registered with the Netherlands Cancer Registry (NCR) – a national database hosted by the Netherlands Comprehensive Cancer Organization (IKNL). However, according to Sabine Siesling, Full Professor at the University of Twente and Senior Researcher at IKNL, pulling together all the necessary information about these 3,500 women was not an easy task. 

Image to the left: Sabine Siesling, Full Professor at the University of Twente and Senior Researcher at IKNL

“In our existing database we were able to find all women diagnosed with breast cancer between 1989 and 2000. But we didn’t have all the information that Gwen Dackus needed about these patients at the start of the project,” says Siesling. “Our data managers, who work in the hospitals, therefore took another look at the patient files of all the selected women to find more specific information about their treatment and wellbeing. These data managers are allowed to check patient files for research purposes and trained to report data in a uniform way, which makes comparing the data of different patients more reliable. Upfront, we had a very good discussion with Gwen about which data were useful to collect and we informed our data managers about the goals of her research project, which is important when you need high quality data.”  

The treatment and outcome data provided for each woman by the NCR was coupled with data from the Netherlands’ PALGA pathology database. The PALGA database contains all digital pathology reports since 1971 that are available for diagnostic and research purposes. Combining the data was done by a so-called ‘trusted third party’, enabling anonymous coupling of privacy-sensitive information from the two databases.  

“As pathologists, we carefully sample from targeted areas of surgical resection specimens to get an overview of all the clinically relevant details of a patient’s tumor,” explains Stefan Willems of UMC Utrecht and vice-chair of PALGA. Filing this overview and archiving the tissue samples is primarily done to help physicians make a diagnosis, during which they can also check older files for the same patient if available. The tissue samples are preserved by fixation with formaldehyde and embedding in blocks of paraffin wax. 

Adding expert pathology to the data 

Having identified the required patient cohort and accumulated the necessary historical data, the next step was to collect all the available tissue samples taken from the women to re-evaluate them, something that involved Dackus traveling from one end of the country to the other. 

In total, around 15 thousand 3-µm thick tissue slices were made from the available tissue blocks, which were then imaged using high-resolution digital pathology slide scanners. The resultant images were stored on an image platform provided by TraIT, one of the initiators of Health-RI, allowing them to be remotely accessed and evaluated by a group of 15 expert pathologists. 

“Because we had digital pathology images, pathologists from Poland and Italy could join the effort too, without the need to send them any physical pathology slides,” says Dackus. “To prevent mistakes, we also made sure that the score sheets that the pathologists used to record their findings were linked automatically to the digital slides of the corresponding patient.” 

Additional tissue slices were cut from the blocks to look for specific receptors in the tumor cells, using colored-marker biochemical analyses that are still possible on formaldehyde-fixed tissue. “We wanted to know which tumors expressed receptors for estrogen or progesterone, two different hormones, and/or receptors for the growth factor HER2, since current treatment decision making is in part based on the presence or absence of these receptors,” explains Dackus. “Compared to women older than 40 diagnosed with breast cancer, young women often lack all these receptors, which reduces their treatment options. For these women we would like to construct a signature that allows you to predict which women need adjuvant treatment and which do not. Next year we hope to be able to present our first results.” 

Paving the way for the future 

Dackus’ research project, which has already been running for five years, is a long-term effort. However, to make this type of research easier, the required research infrastructure still needs to be further professionalized and scaled up – something that Health-RI is determined to continue doing.  

Professor Gerrit Meijer, Chief Scientific Officer of Health-RI and Professor Stefan Willems, PALGA vice-chair

“With Health-RI we want to make life easier for the Gwen Dackusses of the future,” says Professor Gerrit Meijer, Chief Scientific Officer of Health-RI and Head of the Department of Pathology at the Netherlands Cancer Institute.  

 A good example of what needs to be done is the recently introduced PALGA Portal, a joint initiative between PALGA and BBMRI (another co-founder of Health-RI) that makes it much easier for researchers to get the tissue samples they need. 

“In the past, researchers needed to send a separate request to every hospital and these hospitals didn’t always find the time to meet the requests, because patient care is their first priority,” says PALGA vice-chair Willems. “In 2015 we changed the system. Nowadays, researchers can submit their request via the PALGA portal. Each University Medical Centre has a so-called ‘hub employee’, a well-trained technician who is in charge of the collection and distribution of the corresponding tissue blocks in a well-organized manner, saving the researchers a lot of time.” 

Health-RI’s Meijer highlights the fact that the PALGA Portal is not only more efficient but also more GDPR (General Data Protection Regulation) proof. “We had to professionalize the system to improve the protection of patient privacy, and as a next step we want to securely link data from multiple databases,” he says.

“NCR, PALGA and the Hartwig Medical Foundation are already conducting a pilot project together with Health-RI, and if this is successful researchers who want data from these three sources will only have to ask their question once and the accessible data of interest will be returned as a single dataset rather than three independent datasets that researchers need to link together. We’ve received funding from KWF to couple these databases, because they think it’s important to use available patient information for research in an efficient way”.  

IKNL’s Siesling agrees. “We hope our data will be used to improve patient care and quality of life more often. Combining it with biobank data offers researchers many more possibilities. And using our data avoids duplication of work,” she says. However, Meijer admits there are still challenges ahead. “We’re currently looking for funding to sustain this combined portal, because government funding for BBMRI will end in the near future. It will require thinking things over with all stakeholders to consider new costing approaches, such as a service fee that will have to be budgeted for in grant applications. Research funders will begin to appreciate that this is money well spent, because it saves researchers a lot of time and improves the quality of the research output,” he says. 

Another national initiative that will help to facilitate this type of research is the Pathology Image Exchange (PIE) – a collaboration between the Dutch Society for Pathology  (Nederlandse Vereniging Voor Pathologie –NVVP), IKNL and PALGA. “PIE allows the nationwide exchange of digital pathology images, and although primarily developed for diagnostic purposes it is also very interesting for research,” says PALGA vice-chair Willems. “I hope we can start using it for research in cooperation with Health-RI. Furthermore, it will catalyze the use of new techniques like artificial intelligence. By using deep-learning we can probably obtain new interesting information from the digital slides based on things we can’t see with the naked eye. In the end, we hope to have linked clinical, pathology and genetic data that will be exchangeable on a nationwide scale. Imagine the unique wealth of opportunities this will provide for the research community in the Netherlands.” 

This study involved: 

  • Patient records and biobanked tissue samples of 3,500 young breast cancer patients  
  • Expert evaluation of 15,000 tissue slices via high-resolution digital imaging of pathology slides 
  • Biochemical analysis of additional tissue slices to identify specific receptors in tumor cells