Header

UZH-Logo

Maintenance Infos

Modelling Lymphatic Metastatic Progression in Head and Neck Cancer


Ludwig, Roman. Modelling Lymphatic Metastatic Progression in Head and Neck Cancer. 2023, University of Zurich, Faculty of Science.

Abstract

Head and neck squamous cell carcinomas (HNSCCs) are often treated using radiotherapy. Because this type of cancer frequently spreads to regional lymph nodes, parts of the lymphatic system are irradiated alongside the primary tumor. Macroscopic nodal metastasis, visible on imaging modalities such as computed tomography, is always included in the target volume of radiotherapy. Additionally, clinicians electively irradiate lymph node levels (LNLs) when their estimated risk to harbor microscopic disease exceeds a given threshold. Currently, this risk estimate is based on the overall prevalence of metastatic involvement in a given LNL as reported in the literature. However, few published studies on lymphatic metastatic progression patterns investigate how the probability of involvement of LNLs depends on the involvement of other levels and clinicopathological factors such as T-category, which is of crucial importance to create more personalized elective target volumes. This work makes three related contributions to address this problem:

1) We extracted a dataset of patients diagnosed with oropharyngeal SCC at the University Hospital Zurich (USZ). This dataset details the observed involvement per LNL and diagnostic modality together with tumor characteristics including subsite, T-category, lateralization and HPV status. It is thereby the most detailed dataset on patterns of lymphatic progression to date.
2) In addition to sharing the full dataset freely with the research community, we developed an intuitive web page named LyProX. The interface on this web page allows researchers to visually and interactively explore the data extracted at the USZ, as well as another dataset that was kindly provided to us by Vincent Grégoire in response to our efforts.
3) Lastly, we have developed a hidden Markov model (HMM) that predicts the risk of microscopic disease, given an individual patient's clinical diagnosis. It builds on a previous probabilistic model using Bayesian networks (BNs) for the involvement of ipsilateral LNLs I, II, III, and IV, but extends it to naturally include T-category as diagnostic variable. Using Bayesian model comparison, the HMM framework is extended to include the LNLs I, II, III, IV, V, and VII in both the ipsi- and the contralateral side of the neck, accounting for the primary tumor's extension over the mid-sagittal plane as risk factor for contralateral nodal involvement.

Ultimately, we train the HMM with the available data and demonstrate how the resulting predictions may be used to quantitatively predict the personalized risk of occult disease. We find that our model supports a reduction of electively irradiated nodal volumes, especially in the contralateral neck. As such, our model may support the design of future clinical trials on volume-deescalated radiotherapy of HNSCC.

Abstract

Head and neck squamous cell carcinomas (HNSCCs) are often treated using radiotherapy. Because this type of cancer frequently spreads to regional lymph nodes, parts of the lymphatic system are irradiated alongside the primary tumor. Macroscopic nodal metastasis, visible on imaging modalities such as computed tomography, is always included in the target volume of radiotherapy. Additionally, clinicians electively irradiate lymph node levels (LNLs) when their estimated risk to harbor microscopic disease exceeds a given threshold. Currently, this risk estimate is based on the overall prevalence of metastatic involvement in a given LNL as reported in the literature. However, few published studies on lymphatic metastatic progression patterns investigate how the probability of involvement of LNLs depends on the involvement of other levels and clinicopathological factors such as T-category, which is of crucial importance to create more personalized elective target volumes. This work makes three related contributions to address this problem:

1) We extracted a dataset of patients diagnosed with oropharyngeal SCC at the University Hospital Zurich (USZ). This dataset details the observed involvement per LNL and diagnostic modality together with tumor characteristics including subsite, T-category, lateralization and HPV status. It is thereby the most detailed dataset on patterns of lymphatic progression to date.
2) In addition to sharing the full dataset freely with the research community, we developed an intuitive web page named LyProX. The interface on this web page allows researchers to visually and interactively explore the data extracted at the USZ, as well as another dataset that was kindly provided to us by Vincent Grégoire in response to our efforts.
3) Lastly, we have developed a hidden Markov model (HMM) that predicts the risk of microscopic disease, given an individual patient's clinical diagnosis. It builds on a previous probabilistic model using Bayesian networks (BNs) for the involvement of ipsilateral LNLs I, II, III, and IV, but extends it to naturally include T-category as diagnostic variable. Using Bayesian model comparison, the HMM framework is extended to include the LNLs I, II, III, IV, V, and VII in both the ipsi- and the contralateral side of the neck, accounting for the primary tumor's extension over the mid-sagittal plane as risk factor for contralateral nodal involvement.

Ultimately, we train the HMM with the available data and demonstrate how the resulting predictions may be used to quantitatively predict the personalized risk of occult disease. We find that our model supports a reduction of electively irradiated nodal volumes, especially in the contralateral neck. As such, our model may support the design of future clinical trials on volume-deescalated radiotherapy of HNSCC.

Statistics

Downloads

53 downloads since deposited on 17 Feb 2023
49 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Dissertation (monographical)
Referees:Unkelbach Jan, Neupert Titus, Konukoglu Ender, Balermpas Panagiotis, Vogelius Ivan Richter
Communities & Collections:04 Faculty of Medicine > University Hospital Zurich > Clinic for Radiation Oncology
07 Faculty of Science > Physics Institute
UZH Dissertations
Dewey Decimal Classification:530 Physics
Uncontrolled Keywords:head and neck cancer, metastasis, radiotherapy, elective CTV definition, hidden Markov model
Language:English
Place of Publication:Zurich
Date:2023
Deposited On:17 Feb 2023 13:11
Last Modified:17 Feb 2023 13:11
Number of Pages:165
OA Status:Green
Related URLs:https://github.com/rmnldwg/lythesis
https://github.com/rmnldwg/lydata (Research Data)
https://github.com/rmnldwg/lynference (Research Data)
Project Information:
  • Content: Published Version
  • Language: English
  • Licence: Creative Commons: Attribution 4.0 International (CC BY 4.0)