I am an associate professor of biostatistics at the Department of Medical Epidemiology and Biostatistics (MEB) at Karolinska Institutet (KI). I did all my undergraduate studies at Uppsala University, apart from a three-semester stint at the University of Otago in Dunedin, New Zealand, where I studied mathematics and biotechnology. I also did my PhD at Uppsala University, mainly focusing on computer intensive methods for statistical model choice and validation. After I finished my PhD, I did postdocs at the section for Global Safety Assessment at AstraZeneca and at the Department of Medical Epidemiology and Biostatistics, KI. During 2013 and 2014, I was based in San Francisco and worked at the Department of Surgery, University of California San Francisco (UCSF). Since 2015, I am based at MEB where I focus my research on reducing the mortality of breast- and prostate cancer through the use of individualized prevention, diagnostic, and treatment based on the combined use of biomarkers, genomics, imaging, machine learning, and clinical translation via trials.
My research aims to reduce breast- and prostate cancer mortality through the use of individualized prevention, diagnostic, and treatment.
STHLM3 (sthlm3.se) was a large-scale prospective and population-based prostate cancer diagnostic trial involving close to 60,000 men in Stockholm 2013-2015. Two screening methods, PSA and the Stockholm3 test, were assessed and compared for safety and efficacy. The primary aim of the STHLM3 trial was to investigate if the Stockholm3 test could substantially reduce the proportion of men undergoing prostate biopsy whilst maintaining the sensitivity for detecting consequential prostate cancer (Gleason Score 7 or above) compared to PSA. The results showed that the Stockholm3 test could reduce the number of unnecessary prostate biopsies by close to 50% and decrease indolent cancers (Gleason 6, typically regarded as overdiagnosed cancers) by 20%, without reducing the sensitivity to detect consequential prostate cancer compared to using the PSA. The data from the trial provides a fantastic resource for studying prostate cancer diagnostics, and has since the publication of the main results in the Lancet Oncology 2015 been used in a large number of follow-up publications.
We are currently conducting the STHLM3-MRI trials (NCT03377881). That MRI improves prostate cancer diagnostics has become clear over the last few years. During 2016 and 2017, we performed the “STHLM3 MRI Phase 1 – A paired diagnostic trial”, where we included 723 men referred for a prostate biopsy at six centers (4 in Stockholm and 2 in Norway). Stockholm-3 combined with MRI was shown to decrease both unnecessary biopsies and overdiagnosis by about 50% and at the same time increase the sensitivity for consequential prostate cancer (Gleason score ≥ 7) compared to using systematic biopsies on all men. Based on these positive results, we are currently validating them in a large-scale (N=25,000) prospective and population-based randomized screening-by-invitation setting. The aim is to increase the specificity in early detection of prostate cancer without decreasing the sensitivity of aggressive prostate cancers by combining the Stockholm-3 prediction model together with MRI and targeted prostate biopsies. The primary analysis is scheduled for Q3 2020.
AI assisted prostate cancer pathology
Histopathological evaluation of prostate biopsies is critical to the clinical management of men suspected of having prostate cancer. Despite this importance, the histopathological diagnosis of prostate cancer is associated with a number of challenges: The large number of prostate biopsies being performed worldwide together with the shortage of well-trained uro-pathologists and the high inter- pathologist variability leads to suboptimal prostate cancer diagnostics and prognostication, with risks for under- and overtreatment.
We have over the last years developed an artificial intelligence (AI) system to assist the pathologists in the evaluation of prostate biopsies, with the overarching aim of solving these problems. In a recent article in the Lancet Oncology, we demonstrate that the system performs on par with internationally leading uro-pathologist for grading prostate needle biopsies. We are now working on further developing the AI system, in particular by:
- Performing retrospective international multisite validation of the AI system to assess and improve the AI’s performance across different labs, technical platforms, and disease subtypes
- Developing methods to go beyond today’s grading to improve prognostication
- Linking the AI system with genomic profiling of tumor tissue to improve treatment selection
- Performing prospective validation of the AI system in a randomized diagnostic histopathology trial (scheduled to start early 2021).
The WISDOM trial
The WISDOM trial (wisdomstudy.org) investigates whether a risk-based approach to breast cancer screening is as safe and effective as annual mammograms. We will also determine whether women readily accept personalized screening and if knowledge of their own risks, and the reasons for less-frequent screening, will reduce anxiety about breast cancer. Finally, we will determine whether our personalized approach will lead to more of the highest-risk women deciding to use strategies that can prevent breast cancer. Participants in the personalized screening arm receive a risk assessment that includes family and medical history, breast density measurement, and genomic tests (SNPs and high-penetrance genes such as BRCA1/2). WISDOM will enroll 100,000 women across California and South Dakota and is funded by the Patient Centered Research Outcomes Institute (PCORI) and NIH.
Approximately 30% of men diagnosed with prostate cancer develop lethal metastatic castration-resistant prostate cancer (mCRPC). Although recently approved new drugs are beneficial for many mCRPC patients they carry three serious disadvantages. First, these drugs are all very expensive. Second, the response rates to these drugs are low, ranging between 20-40%. This leads to suboptimal treatment and unnecessary side-effects. Third, there are no predictive treatment markers available in clinical care today, which leads to ineffective trial-and-error in treatment decisions. Our hypothesis is that treatment decisions based on molecular profiling will significantly increase progression free survival compared to current clinical care. The vast majority of CRPC metastasize to the bone, with low success rate in obtaining sufficient material. Therefore, we will sequence circulating tumor DNA (ctDNA) consisting of DNA debris from apoptotic cancer being present at high levels in plasma. We are currently testing this hypothesis in the ProBio trial starting in the end of 2018, a large, international, multicentre, randomized study in men with mCRPC. ProBio (NCT03903835) uses an outcome-adaptive randomization trial design, with the goal to learn as rapidly as possible which treatments are effective for which ctDNA biomarker profiles. We are currently working on extending the ProBio platform trial to also encompass men with metastatic hormone sensitive prostate cancer (mHSPC).
Current group members
- Hari Vigneswaran (PhD student)
- Joel Andersson (PhD student)
- Kimmo Kartasalo (postdoc)
- Andreas Karlsson (postdoc)
- Henrik Olsson (PhD student)
- Alessio Crippa (postdoc)
- Andrea Discacciati (statistician)
I am currently a co-supervisor of eight PhD students.
Previous PhD students and postdocs
- Thorgerdur Palsdottir (PhD student, graduated 2019)
- Peter Ström (PhD student, graduated 2020)
- Anna Lantz (postdoc, now independent PI)
Current research grants (as PI)
My research is supported by the Swedish Research Council (Vetenskapsrådet), the Swedish Research Council for Health, Working Life and Welfare (FORTE), the Swedish Cancer Society (Cancerfonden), Karolinska Institutet, the European Institute of Innovation & Technology (EIT), Prostatacancerförbundet, Åke Wiberg Foundation, the Stockholm County Council (SLL), the Patient Cnetered Research Outcome Institutet (PCORI), and the National Institutet for Health (NIH).
I currently teach the course “Multivariable prediction modeling with applications in precision medicine” (LK2990) together with Mattias Rantalainen.
- P. Ström, K. Kartasalo, H. Olsson, L. Solorzano, B. Delahunt, D. Berney, D. Bostwick, A. Evans, P. Humphrey, K. Iczkowski, J. Kench, G. Kristiansen, T. van der Kwast, K. Leite, J. McKenney, J. Oxley, C. Pan, H. Samaratunga, J. Srigley, H. Takahashi, T. Tsuzuki, M. Varma, M. Zhou, J. Lindberg, C. Lindskog, P. Ruusuvuori, C. Wählby, H. Grönberg, M. Rantalainen, L. Egevad, and M. Eklund. Artificial intelligence for diagnosis and grading of prostate cancer in biopsies: a population-based, diagnostic study. Lancet Oncol. 2020 Feb;21(2):222-232. PMID 31926806.
- A. Wallerstedt, P. Ström, H. Grönberg, T. Nordström, and M. Eklund. Risk of prostate cancer in men treated with 5α-reductase inhibitors – results from a large population-based prospective study.Journal of the National Cancer Institute, 2018, 110(11):1216-1221. PMID: 29548030.
- H. Grönberg, M. Eklund, W. Picker, M. Aly, F. Jäderling, J. Adolfsson, M. Landquist, E. Skaaheim Haug, P. Ström, S. Carlsson, and T. Nordström. Detection of prostate cancer using a combination of blood- based risk prediction and multiparametric magnetic resonance imaging. European Urology, 2018, 74(6):722-728. PMID: 30001824.
- P. Ström, T. Nordström, H. Grönberg, and M. Eklund. The Stockholm-3 model (S3M) for prostate cancer detection: biomarker contribution and reflex test potential. European Urology, 2018, 74(2), 204-210. PMID: 29331214.
- M. Eklund, K. Broglio, J. Connor, C. Yau, and L. Esserman. The WISDOM Personalized Breast Cancer Screening Trial: Simulation Study to Assess Potential Bias and Analytic Approaches. JNCI: Journal of the National Cancer Institute Cancer Spectrum, 2019, 2 (4), pky067. PMID: 31360882.
- Y. Shieh, M. Eklund, L. Madlensky, S. Sawyer, C.K. Thompson, A. Stover Fiscalini, E. Ziv, L.J. van't Veer, L.J. Esserman, and J.A. Tice. Breast Cancer Screening in the Precision Medicine Era: Risk-Based Screening in a Population-Based Trial. JNCI: Journal of the National Cancer Institute, 2017, 109:5, djw290. PMID: 28130475.
- T. Nordström, J. Adolfsson, H. Grönberg, M. Eklund. Repeat Prostate-Specific Antigen Tests Before Prostate Biopsy Decisions. JNCI: Journal of the National Cancer Institute. 2016 Jul 14;108(12). pii: djw165. doi: 10.1093/jnci/djw165. PMID: 27418620.
- Shieh, M. Eklund, G.F. Sawaya, W.C. Black, B.S. Kramer, and L. Esserman. Population cancer screening: hope and hype. Nature Reviews Clinical Oncology, 2016, 13:9, 550-65. PMID: 27071351.
- H. Grönberg, J. Adolfsson, M. Aly, T. Nordström, P. Wiklund, Y. Brandberg, J. Thompson, F. Wiklund, J. Lindberg, M. Clements, L. Egevad, and M. Eklund. Prostate cancer screening in men aged 50-69 years (STHLM3): a prospective population-based diagnostic study. Lancet Oncology, 2015, 16:16, 1667-76. PMID: 26563502.
- T. Nordström, A. Vickers, M. Assel, H. Lilja, H. Grönberg, and M. Eklund. Comparison between the four-kallikrein panel and Prostate Health Index (PHI) for predicting prostate cancer. European Urology, 2015, 68:1, 139-46. PMID: 25151013.
For a list of medical papers I have co-authored, please see this link to PubMed.
- PhD Bioinformatics, Uppsala University, 2010
- MSc Molecular Biotechnology (civilingenjör), Uppsala University, 2005
- MSc Mathematics, Uppsala University, 2006
- BSc Macroeconomics, Uppsala University, 2004