STHLM3 Model Improves High-Risk Prostate Cancer Screening Specificity

The Stockholm 3 model improves screening specificity for high-risk prostate cancer without loss of sensitivity

medwireNews: Results from the Stockholm 3 (STHLM3) study show that combining plasma biomarkers, genetic polymorphisms and clinical characteristics can improve specificity for high-risk prostate cancer detection in men aged 50–69 years compared with prostate-specific antigen (PSA) testing.

“Use of the STHLM3 model in structured screening could reduce the number of prostate biopsy samples taken by about a third compared with the use of PSA screening”, say lead author Henrik Grönberg, from the Karolinska Institutet in Stockholm, Sweden, and co-investigators.

“Importantly, this can be achieved without compromising the number of high risk cancers diagnosed”, they report in The Lancet Oncology.

The STHLM3 screening model involves measuring plasma levels of PSA, free PSA, intact PSA, hexokinase 2, beta-microseminoprotein and macrophage inhibitory cytokine 1, and genotyping patients for 232 single nucleotide polymorphisms, the researchers explain. Patient age, family history and previous prostate biopsy and examinations are also considered.

The model was initially applied to a training cohort of 11,130 men and then to a validation cohort of 47,688 men, all of whom were aged 50–69 years and free from prostate cancer.

In all, 7606 men in the validation cohort were referred for urological consultation and underwent biopsy on the basis of a PSA greater than 3 ng/mL or a STHLM3 cutoff score indicating a Gleason score of at least 7.

In multivariate analysis, all of the variables in the STHLM3 score were significantly associated with high-risk disease and the overall score was significantly better than PSA for detection of high-risk prostate cancers, with an area under the curve of 0.74 versus 0.56.

When used at the same sensitivity as the PSA test, corresponding to 10% predicted risk, the STHLM3 model would reduce the number of biopsies by 32% and the number of benign biopsies by 44%, the researchers report.

And 21% of the 603 high-risk prostate cancers detected by STHLM3 were in men with a PSA of 1–3 ng/mL.

The number of Gleason score 6 prostate cancers identified as high risk was reduced by 17% with the STHLM3 model , from 867 detected by PSA alone to 722. The 104 Gleason score 6 tumours not detected by STHLM3 all had a total cancer length of below 10 mm in biopsy “indicating that most of these cancers were clinically insignificant”, the team writes.

Discussing the findings in an accompanying comment, Alastair Lamb and Ola Bratt, from Addenbrooke’s Hospital in Cambridge, UK, say that the STHLM3 study is “proof-of-concept for increasing the specificity of PSA-based prostate cancer screening at a population level.”

However, they say the results do not yet shift the balance of benefit versus harm for population-based screening, noting that the model has not been tested for repeat screening or demonstrated to show an impact on mortality.

And they emphasise that the issue of overdiagnosis has not been resolved, with over half of detected cancers of Gleason score 6 or 7 (3+4) and likely to be clinically insignificant.

“Presently we do not have the means to accurately predict which low grade prostate cancers will progress, but genetic profiling methods look promising”, they conclude.


Grönberg H, Adolfsson J, Aly M, et al. Prostate cancer screening in men aged 50–69 years (STHLM3): a prospective population-based diagnostic study. Lancet Oncol 2015; Advance online publication 9 November. DOI:

Lamb AD, Bratt O. Towards “next generation” prostate cancer screening. Lancet Oncol 2015; Advance online publication 9 November. DOI:

medwireNews ( ) is an independent medical news service provided by Springer Healthcare Limited. © Springer Healthcare Ltd; 2015