Eva Onjukka

Eva Onjukka

Affiliated to Research
Visiting address: M1:01 Anna Steckséns väg 41, 17176 Stockholm
Postal address: K7 Onkologi-Patologi, K7 Forskning Tedgren, 171 77 Stockholm

About me

  • * Medical physicist at Karolinska University Hospital and R&
  • D co-ordinator
    for external-beam radiotherapy physics
    * Fellowship from the Swedish Cancer Society: Real-world data for
    evidence-based radiotherapy treatment planning for head and neck cancer
    * PhD at the University of Liverpool, UK, in 2011
    * Active member of European Society for Radiotherapy and Oncology (ESTRO)
    * Research interests: Radiobiological modelling, Radiotherapy side effects,
    optimization of radiotherapy treatment plans
  • primarily within the areas
    of H&
  • N cancer and SBRT
    * Supervisor experience: 1 completed PhD project, 2 ongoing, 7 completed
    Master’s projects
    * Also published under the name of Eva Rutkowska


  • Every year around 4000 patients are treated with radiotherapy at Karolinska
    University hospital. Most patients receive external-beam radiotherapy but the
    hospital also offers treatments with brachytherapy, radiosurgery and proton
    therapy. The treatments are optimised by a team of medical physicists,
    oncologists and nurses and we have many projects aiming to learn from the
    result of different treatments and continuously improve treatment methods.
    Since every patient has a unique anatomy and each tumour to be treated
    presents differently, an optimised treatment results in very different dose
    distributions for each patient. The probability of side effects from
    radiotherapy depends strongly on the 3D dose-distribution in each organ at
    risk. Detailed information about the treatment plans is saved in the
    treatment planning system, including the full dose distribution in each part
    of the patient.
    -------- Projects in head &
  • neck cancer --------------------------------------
    Since 2013, the follow-up data of all head &
  • neck cancer patients are entered
    into an electronic quality registry, including side-effect data. Currently
    the registry contains over a thousand patients, sometimes with several years
    of follow-up after radiotherapy. In combination with the treatment data,
    these offer many opportunities to develop models of the effect of the
    treatment in order to systematically learn from past experience. My work is
    supported by a fellowship from the Swedish Cancer Society and a grant from
    Varian Medical Systems. Previous work has been funded by grants from the
    Cancer Research Funds of Radiumhemmet.
    .... Completed research
    We have studied the outcome for head &
  • neck cancer patients who received
    reirradiation. We introduced a clear definition of reirradiation where the
    overlapping volume received at least 60 Gy from the first treatment and at
    least 40 Gy at reirradiation. The volume receiving at least 100 Gy correlated
    with the risk of acute side effects. Dose constraints for carotid blow-out
    and osteoradionecrosis from the literature could be confirmed with our data,
    where the 3D dose distributions were summed after deformable registration
    between the CT images used for treatment planning. The follow-up showed
    long-term survival without serious side effects after reirradiation for many
    patients, but also that with long-term follow-up a higher incidence of
    serious side effects is observed compared to previous studies with shorter
    A retrospective analysis of a cohort treated with dose escalation showed
    equivalent results for boost with brachytherapy and simultaneous integrated
    boost with VMAT. Dose-escalated treatments were compared with standard
    treatments but the conclusions were limited by a skewed distribution between
    the groups with respect to important prognostic factors.
    The research in reirradiation and dose escalation comprised Anna Embring's
    PhD projects.
    .... Current research
    We have a vision to develop a framework for evidence-based treatment
    planning. This includes developing a technical solution for the integration
    of predictive models in the treatment planning system, with support for both
    simple and complex dose metrics. But it also includes developing a strategy
    for identifying the most appropriate evidence to rely on in the design of the
    treatment plans. The strength of our locally developed models is that they
    are based on registry data (so-called real-world data), including cases of
    very long follow-up, where we can expect the training data to match our
    future patients well. On the other hand, there are published models on more
    controlled datasets, which may suffer less from unintended bias, and which
    have been validated in an external cohort.
    Meanwhile, work continues to analyse the risk of side effects based on the
    local quality registry. A multivariate model of the risk of developing
    xerostomia after radiotherapy of head &
  • neck cancer has been developed based
    on 753 patients. As the model is based on registry data (real-life data)
    rather than data from clinical studies, as is more common in the literature,
    the cohort is unusually large and models suitable for decision support can be
    developed. The results have been published and the models will be further
    refined and adapted for clinical use.
    We are also investigating the dose response for late dysphagia after
    radiotherapy. A voxel-based analysis aiming to identify which parts of the
    anatomy which, if damaged from radiation, can cause swallowing dysfunction is
    ongoing. I’m developing a routine for automatic registration of CT images
    to a standard anatomy in order to detect in which areas the radiation dose
    differs between patients with and without swallowing dysfunction,
    respectively. Using an alternative approach, different anatomical structures
    are delineated (segmented) and dose statistics from these structures can be
    compared to the outcome. Here, the possibility of using AI-based automatic
    segmentation of the structures of interest is explored.
    -------- Projects in SBRT ----------------------------------------------------
    Since SBRT (stereotactic body radiotherapy) was pioneered at Karolinska
    University in the early 1990’s, many patients with many different diagnoses
    have been treated with this technique. I participate in several projects
    aiming to evaluate the effect of the treatment, related to the unique dose
    distribution delivered to each patient.
    .... Current research
    HILUS is a multi-centre study of the effect of SBRT for patients with
    centrally located lung tumours. Several cases of acute bleedings were
    observed and it is important to discover which aspects of the treatment
    and/or the disease which result in this side effect. In order to determine
    whether the radiation dose to the main bronchi affects the risk of acute
    bleeding a larger, retrospective, cohort was included in the next phase of
    the study (HILUS III). The first analysis was published in July 2023, and a
    DVH-based dose response analysis is ongoing. This study highlighted high dose
    to the main bronchi and the intermediate bronchus as particularly risky. In
    2022, we published a review article on the sensitivity to dose to different
    parts of the bronchial tree.
    In HILUS IV, bronchi have been delineated in CBCT- and 4DCT images to study
    the uncertainty in delivered dose to the bronchi. Then the accumulated dose
    will be calculated with the intention of creating a stronger dose-response
    .... Completed research
    When apical lung tumours are treated with SBRT, the brachial plexus may
    receive a high dose which sometimes leads to brachial plexopathy. In 2019 we
    published the incidence in a cohort of 52 retrospectively analysed patients,
    including a dose response model.
    The effect of SBRT for patients with large tumours (>
  • 5 cm in diameter) has
    been analysed retrospectively in a large cohort of patients treated for
    different diagnoses and tumour locations in the thorax and abdomen. A
    predictive model of local control has been developed, including dose/volume
    parameters of the tumour.


All other publications

News from KI

Events from KI