Article Text
PDF
Abstract
Diagnosing cystic fibrosis (CF) when sweat chloride is not in the CF range and less than 2 disease-causing CFTR mutations are found requires physiological CFTR assays, which are not always feasible or available. We developed a new physiological CFTR assay based on the morphological differences between rectal organoids from subjects with and without CF. In organoids from 167 subjects with and 22 without CF, two parameters derived from a semi-automated image analysis protocol (rectal organoid morphology analysis, ROMA) fully discriminated CF subjects with two disease-causing mutations from non-CF subjects (p<0.001). ROMA, feasible at all ages, can be centralised to improve standardisation.
- cystic fibrosis
Statistics from Altmetric.com
Introduction
The diagnosis of cystic fibrosis (CF), an autosomal recessive disease caused by CF transmembrane conductance regulator (CFTR) gene mutations, relies on abnormal (≥60 mmol/L) sweat chloride concentration (SCC) and/or two disease-causing CFTR mutations, as defined by the CFTR2 database.1 2 In some subjects either with CF compatible symptoms or after neonatal screening, the diagnosis cannot be confirmed nor excluded. For patients with only intermediately elevated (30–60 mmol/L) SCC and CFTR mutation(s) of varying or unknown clinical consequence, second-line diagnostic tests (nasal potential difference (NPD) and intestinal current measurements (ICM)) are advocated to further explore CFTR function and assist the diagnosis. These tests are not readily available nor feasible at all ages.1
Rectal organoids are 3D structures grown from intestinal stem cells in a mucosal sample obtained through rectal biopsy. CFTR protein expression is maintained in and determines the morphology of these organoids, inducing swelling of non-CF organoids through salt and water accumulation in the lumen surrounded by a cellular layer, while CF organoids have no lumen as described by the Beekman group.3 4
We quantified morphological differences to explore the ability of rectal organoid morphology analysis (ROMA) to differentiate organoids from subjects with a clear-cut diagnosis of CF from those of subjects without CF, giving proof of concept for ROMA as a diagnostic test for CF.
Methods
Organoids from a convenience cohort of 212 subjects were collected and imaged by one researcher blinded to subject characteristics. Twenty-three subjects were excluded due to low-quality images. Organoids of 167 subjects with CF and two disease-causing CFTR mutations2 and 22 non-CF subjects were analysed (online supplemental file 1 and 2.1).
Supplemental material
Rectal biopsies and organoid cultures were performed as previously described (online supplemental file 2.2).4 No adverse events were reported. In total, 32 wells per subject were plated under basal conditions (no CFTR modulators nor forskolin added) and images of each well were acquired after overnight growth and calcein staining.
Two indices were calculated using imaging software (NIS-Elements AR Analysis 5.02.00) to quantify organoid morphology: the intensity ratio (IR) measures the presence or absence of a central lumen, and the circularity index (CI) quantifies the roundness of the organoids (figure 1 and online supplemental file 2.3).
Images of rectal organoids from subjects without (upper panels) and with CF (lower panels). Illustration of the methods to calculate the two indexes, IR (intensity ratio; central panel) and CI (circularity index; right panel), used to quantify morphological differences between rectal organoids of subjects with and without CF. IR measures the presence or absence of a central lumen, calculated in three steps: (1) calculate the global fluorescence intensity of the organoids: erode 1 pixel (2.5 µm) to remove the surrounding ‘halo’ around each structure and measure the mean fluorescence intensity of the remaining whole organoid; (2) calculate the lumen fluorescence intensity of the organoids: erode 10 pixels from each structure to remove the cellular border from the organoids and measure the mean fluorescence intensity of the remaining structure; (3) IR is equal to (intensity central part of organoid)/(intensity whole organoid) and is higher in CF than in non-CF organoids. CI quantifies the roundness of the organoids, defined as (4×π×area)/perimeter, which is lower in CF than in non-CF organoids. CF, cystic fibrosis; IR, intensity ratio; CI, circularity index.
Discriminant analysis was applied to differentiate between CF and non-CF subjects using IR and CI. Mann-Whitney and Fisher exact tests were used for between-group comparisons.
Results
The IR and CI differed (p<0.001) between CF and non-CF (table 1). Perfect discrimination (AUC=1) was obtained with a linear discriminant analysis (figure 2). Analysis of only eight wells chosen randomly out of the 32 showed almost identical results (AUC=1) (online supplemental file 2.4).
Intensity ratio (IR) and circularity index (CI) values of each subject according to disease status, pancreatic status and sweat chloride concentration. The line represents the optimal discrimination line obtained by linear discriminant analysis. CF, cystic fibrosis; PI, pancreatic insufficient; PS, pancreatic sufficient; SCC, sweat chloride concentration.
Baseline characteristics of the subjects and indexes calculated using rectal organoid morphology analysis (ROMA)
No correlation between ROMA indexes and age was found, nor differences in IR and CI between subjects with CF in the lowest versus three highest SCC quartiles (<87 and ≥87 mmol/L, respectively). The IR (p=0.007), but not CI (p=0.419), was different between PI (pancreatic insufficient) and PS (pancreatic sufficient) subjects with CF. For additional statistical analyses, see online supplemental files 2.4 and 3.
Discussion
Both IR and CI, calculated with ROMA, discriminated subjects with clinical CF and two disease-causing CFTR mutations from non-CF subjects. This makes ROMA appealing as an additional physiological CFTR assay to assist in the diagnosis of CF, with a discriminative ability comparable with that reported for SCC and other physiological CFTR assays such as NPD or ICM.5 6
Similarly to SCC,5 mean IR is different between PS and PI subjects with CF, although not fully discriminative. Residual CFTR function, translating in pancreatic sufficiency, leads to the presence of a small lumen in the organoids and thus lower IR, without altering the organoids’ irregular shape quantified by CI.
The current study is monocentric and needs replication, but samples have already been successfully received from other Belgian and international centres. Before ROMA can be proposed as diagnostic test for CF, more data are needed, especially from patients with a clinical CF diagnosis but without two disease-causing CFTR mutations, and from prospective analysis in subjects with unresolved CF diagnosis, including infants with CF screening positive, inconclusive diagnosis (CFSPID). The absence of a gold standard will need to be taken into account, as with other CFTR physiological assays.
Rectal suction biopsy for ROMA can easily be performed in general hospitals with low complication rates,7 even in infants (with CFSPID). Quality requirements for biopsies are lower for ROMA8 than for ICM,9 as only the presence of viable crypts is needed to grow an organoid culture, rather than an intact epithelium required for electrophysiology. Biopsies can thus be shipped to a central laboratory for analysis,8 which ensures standardisation together with the semi-automated nature of the analysis.
Possible issues with ROMA include the complexity and cost of organoid culture. However, standardised protocols are available for organoid culture and technique portability has been demonstrated.8 Reducing the number of wells from 32 to 8 resulted in virtually identical results and would make ROMA more cost-efficient. Using rectal instead of airway tissue is both an advantage (CFTR specific, no influence of disease state)1 and a disadvantage (possible differences in CFTR function between organs). The delay of 1–2 months between taking the biopsy and availability of results is rarely an issue as these cases are often not urgent. This waiting time is similar to that of extensive genetic analysis.
Further work will have to assess ROMA as a diagnostic test in patients with equivocal diagnosis. Beyond diagnosis, organoids used for ROMA can be biobanked for later personalised medicine research for the newly diagnosed patients with CF.10 ROMA could also contribute to measuring modulator efficiency, as restoring CFTR function to high levels causes the appearance, before any stimulation, of a central lumen and swelling to a more round shape,3 4 measured by IR and CI, respectively.
Ethics statements
Ethics approval
Approval from the Ethical committee of UZ Leuven and informed consent from all subjects or representatives were obtained.
Acknowledgments
We thank the patients and parents who participated in this study. We thank Els Aertgeerts, Karolien Bruneel, Claire Collard, Liliane Collignon, Monique Delfosse, Anja Delporte, Nathalie Feyaerts, Cécile Lambremont, Lut Nieuwborg, Nathalie Peeters, Ann Raman, Pim Sansen, Hilde Stevens, Marianne Schulte, Els Van Ransbeeck, Christel Van de Brande, Greet Van den Eynde, Marleen Vanderkerken, Inge Van Dijck, Audrey Wagener, Monika Waskiewicz and Bernard Wenderickx for logistic support. We also thank Stefan Joris and Dr. Jan Vanleeuwe from the Mucovereniging/Association Muco for their support.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Collaborators Belgian Organoid Project: Hedwige Boboli (CHR Citadelle, Liège, Belgium), Linda Boulanger (University Hospital Leuven, Belgium), Georges Casimir (HUDERF, Brussels, Belgium), Benedicte De Meyere (University Hospital Ghent, Belgium), Elke De Wachter (University Hospital Brussels, Belgium), Danny De Looze (University Hospital Ghent, Belgium), Isabelle Etienne (CHU Erasme, Brussels, Belgium), Laurence Hanssens (HUDERF, Brussels), Christiane Knoop (CHU Erasme, Brussels, Belgium), Monique Lequesne (University Hospital Antwerp, Belgium), Vicky Nowé (GZA St. Vincentius Hospital Antwerp), Dirk Staessen (GZA St. Vincentius Hospital Antwerp), Stephanie Van Biervliet (University Hospital Ghent, Belgium), Eva Van Braeckel (University Hospital Ghent, Belgium), Kim Van Hoorenbeeck (University Hospital Antwerp, Belgium), Eef Vanderhelst (University Hospital Brussels, Belgium), Stijn Verhulst (University Hospital Antwerp, Belgium), Stefanie Vincken (University Hospital Brussels, Belgium).
Contributors Conceptualisation: SC, ASR, FV and KDB. Methodology: SC, ASR, NC, SM, KDB and FV. Recruiting of CF subjects and collection of rectal biopsies: MB, MP, FV and LD. Recruiting of control subjects and collection of rectal biopsies: MB, MP, FV, LD, KA and MF. Culturing the organoids and performing the rectal organoid morphology analysis: SC, EF and ASR. Analysis of the results and figures preparation: SC, ASR, SF and FV. Writing—original draft: SC, ASR and FV. Writing—review and editing: SC, ASR, NC, SF, EF, KA, MF, CV, SM, MB, MP, LD, KDB, FV, HB, LB, GC, BDM, EDW, DDL, IE, LH, CK, ML, VN, DS, SVB, EVB, KVH, EV, SVe, SVi. Supervision: ASR, FV and KDB.
Funding This study was funded by the Belgian CF patients’ association “Mucovereniging/Association Muco”, the Research Grant of the Belgian Society of Paediatrics BVK-SBP 2019, and a grant from the UZ Leuven Fund for Translational Biomedical Research.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.