MR Biomarkers and The Kaiser Score

The cancer-typical “proliferation – vascularization – hypoxia” cycle is accessible by imaging. Pascal Baltzer explained which method can be used for which type of breast cancer. The new “Kaiser Score” might help to put it all together.

Breast cancer is a heterogeneous disease. Radiology is currently sorting out how imaging markers might help in clustering this heterogeneity.

Pascal Baltzer, Vienna University, Austria, showed that each pathological tissue change in the “proliferation – vascularization – hypoxia” cycle has biomarkers that can be matched with breast cancer phenotypes.

Proliferation – Hypoxia – Vascularization

First of all, regular MRI morphology and growth patterns provide basic information on tumor proliferation. Other methods add value:

  • 18FDG-PET is able to measure glucose metabolism.
  • 1H-MR Spectroscopy gives information on choline metabolism and water content.
  • Diffusion weighted imaging (DWI) metrics give insight into tumor microanatomy.

Hypoxia can be measured using

  • 18FMISO-PET, which uses [F-18] fluoromisonidazole as the hypoxid tracer, or
  • qBOLD (quantitative BOLD), which provides a regional in vivo oxygen extraction fraction (OEF) measurement, or
  • MR Spectroscopy (MRS), which is sensitive to changes in lactate concentration, which is often due to hypoxia.

Vascularization is shown via dynamic contrast enhanced (DCE) MRI and vessel architecture mapping.

The absence of enhancement in DCE MRI is equal to no biologically relevant cancer, said Baltzer – whereas contrast enhancement speaks for tumor angiogenesis of more than 2-3mm. “The more washout, the more malignant the tumor is”, he added as a rule-of-thumb.

DCE MRI also predicts the risk for distant metastases and disease-specific survival. Volumetric enhancement analysis is now taking this approach even further.

Additional Options

 “Spectroscopy is probably not dead,” said Baltzer. MRS is able to assess cellular stress metabolism. It also allows cancer subtyping in vivo. The combination of MRS and PET allows identifying very aggressive phenotypes.

A synergistic multiparametric PET-MRI approach further specifies tumor dignity. There are various PET-MRI markers for tissue vitality, metastatic potential and drug delivery.

The Kaiser Score Prevents Unnecessary Biopsies

“Our main issue is both the lack of data and a data overload,” argued Baltzer. “There is lots of data and we do not quite know what it is worth,” he added. He identified machine learning as a viable option to separate the good from the bad.

The Kaiser Score – named after one of the breast MRI founding parents Werner Kaiser – is a first major step showing what machine learning applied to BI-RADS can do: The algorithm is based on more than 1000 histologically verified lesions, all of which were independently validated (Dietzel 2018).

It combines five independent diagnostic BI-RADS lexicon criteria (margins, SI-time curve type, internal enhancement and presence of edema) in one flowchart. The resulting score then specifies probabilities of malignancy, which can be used for evidence-based decision-making.
“The Kaiser score may save up to one forth of unnecessary breast biopsies,” explained Baltzer. It has also shown to reduce inter-reader variation, and improves less experienced reader performance.

Baltzer believes there needs to be more exploitable data. “To me, this is the most pressing problem – let’s start to collaborate”, he said.


Dietzel M, Baltzer PT. How to use the Kaiser score as a clinical decision rule for diagnosis in multiparametric breast MRI: a pictorial essay. Insights Imaging. 2018; 9(3): 325–35.

ECR 2019
AutorIn: kf/ktg,
Related content: AKI, CT

Presentation Title: Non-invasive phenotyping of breast cancer with novel MR biomarkers
Speaker: Pascal Baltzer, Vienna University, Austria
Date: Wednesday, 27th February 2019
Session Code: SY2b