Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Kidney function and shrunken pore syndrome - epidemiological results and methodological issues

Åkesson, Anna LU orcid (2026) In Lund University, Faculty of Medicine Doctoral Dissertation Series
Abstract
Background: Glomerular filtration rate (GFR) is a key indicator of kidney function, typically estimated using creatinine or cystatin C. When cystatin-C-based eGFR is markedly lower than creatinine-based eGFR, this discrepancy may reflect a selective impairment in the filtration of medium-sized molecules, referred to as Shrunken Pore Syndrome (SPS), or more broadly, Selective Glomerular Hypofiltration Syndromes (SGHS). SPS/SGHS has been linked to increased morbidity and mortality, but its clinical significance, optimal diagnostic definition, and methodological implications remain insufficiently understood.
Objectives: This thesis aimed to evaluate the prognostic value of SPS/SGHS in clinical and population-based cohorts... (More)
Background: Glomerular filtration rate (GFR) is a key indicator of kidney function, typically estimated using creatinine or cystatin C. When cystatin-C-based eGFR is markedly lower than creatinine-based eGFR, this discrepancy may reflect a selective impairment in the filtration of medium-sized molecules, referred to as Shrunken Pore Syndrome (SPS), or more broadly, Selective Glomerular Hypofiltration Syndromes (SGHS). SPS/SGHS has been linked to increased morbidity and mortality, but its clinical significance, optimal diagnostic definition, and methodological implications remain insufficiently understood.
Objectives: This thesis aimed to evaluate the prognostic value of SPS/SGHS in clinical and population-based cohorts (studies I and II); assess the diagnostic performance of eGFRratio versus eGFRdifference for identifying SGHS (study III); investigate whether machine-learning models can improve GFR estimation compared with established equations (study IV); and examine whether past eGFR values enhance current GFR estimation accuracy (study V).
Methods: Analytical frameworks included Cox regression, generalized propensity scores, novel quartet-based matching, comparison of eGFRratio versus eGFRdifference, and evaluation of machine-learning models against the EKFC equation. Additionally, innovative weighted and Bayesian averaging strategies were developed to integrate past eGFR values.
Results: SPS/SGHS was prevalent in both clinical (23%) and population-based (8%) cohorts and strongly associated with higher all-cause mortality, independent of measured GFR, comorbidities, or demographic factors. Both eGFRratio and eGFRdifference predicted mortality with similar discriminatory performance, although they classified partially distinct subgroups. Machine-learning models did not outperform the EKFC equation, and differences in accuracy were clinically negligible. Incorporating previous eGFR values provided modest yet consistent improvements in predictive performance, particularly when restricted to narrow time‑windows that reflect short‑term physiological stability.
Conclusions: Selective impairments in filtration quality represent an important but under-recognized dimension of kidney dysfunction. SPS/SGHS is a robust predictor of adverse outcomes, and combined use of eGFRratio and eGFRdifference may improve risk stratification. Advances in modeling appear limited by the informational content of current biomarkers. Incorporating past eGFR values into current GFR estimation produces modest but reproducible improvements, particularly when restricted to short time-windows.
(Less)
Please use this url to cite or link to this publication:
author
supervisor
opponent
  • Professor Theodorsson, Elvar, Linköping University, Linköping, Sweden
organization
publishing date
type
Thesis
publication status
published
subject
keywords
Glomerular filtration rate (GFR), Shrunken pore syndrome, Creatinine, Cystatin C, Propensity scores, Machine learning
in
Lund University, Faculty of Medicine Doctoral Dissertation Series
issue
2026:52
pages
74 pages
publisher
Lund University, Faculty of Medicine
defense location
Belfragesalen, BMC D15, Klinikgatan 32 i Lund
defense date
2026-04-17 09:00:00
ISSN
1652-8220
ISBN
978-91-8021-850-4
language
English
LU publication?
yes
id
fc671591-a645-4f85-afe4-9954cfc265ba
date added to LUP
2026-03-09 13:29:48
date last changed
2026-03-13 09:08:27
@phdthesis{fc671591-a645-4f85-afe4-9954cfc265ba,
  abstract     = {{<b>Background:</b> Glomerular filtration rate (GFR) is a key indicator of kidney function, typically estimated using creatinine or cystatin C. When cystatin-C-based eGFR is markedly lower than creatinine-based eGFR, this discrepancy may reflect a selective impairment in the filtration of medium-sized molecules, referred to as Shrunken Pore Syndrome (SPS), or more broadly, Selective Glomerular Hypofiltration Syndromes (SGHS). SPS/SGHS has been linked to increased morbidity and mortality, but its clinical significance, optimal diagnostic definition, and methodological implications remain insufficiently understood.<br/><b>Objectives: </b>This thesis aimed to evaluate the prognostic value of SPS/SGHS in clinical and population-based cohorts (studies I and II); assess the diagnostic performance of eGFR<sub>ratio </sub>versus eGFR<sub>difference </sub>for identifying SGHS (study III); investigate whether machine-learning models can improve GFR estimation compared with established equations (study IV); and examine whether past eGFR values enhance current GFR estimation accuracy (study V).<br/><b>Methods:</b> Analytical frameworks included Cox regression, generalized propensity scores, novel quartet-based matching, comparison of eGFR<sub>ratio</sub> versus eGFR<sub>difference</sub>, and evaluation of machine-learning models against the EKFC equation. Additionally, innovative weighted and Bayesian averaging strategies were developed to integrate past eGFR values.<br/><b>Results:</b> SPS/SGHS was prevalent in both clinical (23%) and population-based (8%) cohorts and strongly associated with higher all-cause mortality, independent of measured GFR, comorbidities, or demographic factors. Both eGFR<sub>ratio</sub> and eGFR<sub>difference</sub> predicted mortality with similar discriminatory performance, although they classified partially distinct subgroups. Machine-learning models did not outperform the EKFC equation, and differences in accuracy were clinically negligible. Incorporating previous eGFR values provided modest yet consistent improvements in predictive performance, particularly when restricted to narrow time‑windows that reflect short‑term physiological stability.<br/><b>Conclusions:</b> Selective impairments in filtration quality represent an important but under-recognized dimension of kidney dysfunction. SPS/SGHS is a robust predictor of adverse outcomes, and combined use of eGFR<sub>ratio</sub> and eGFR<sub>difference</sub> may improve risk stratification. Advances in modeling appear limited by the informational content of current biomarkers. Incorporating past eGFR values into current GFR estimation produces modest but reproducible improvements, particularly when restricted to short time-windows.<br/>}},
  author       = {{Åkesson, Anna}},
  isbn         = {{978-91-8021-850-4}},
  issn         = {{1652-8220}},
  keywords     = {{Glomerular filtration rate (GFR); Shrunken pore syndrome; Creatinine; Cystatin C; Propensity scores; Machine learning}},
  language     = {{eng}},
  number       = {{2026:52}},
  publisher    = {{Lund University, Faculty of Medicine}},
  school       = {{Lund University}},
  series       = {{Lund University, Faculty of Medicine Doctoral Dissertation Series}},
  title        = {{Kidney function and shrunken pore syndrome - epidemiological results and methodological issues}},
  url          = {{https://lup.lub.lu.se/search/files/244432278/Avhandling_Anna_kesson_LUCRIS.pdf}},
  year         = {{2026}},
}