Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes

May Yee Choi; Irene Chen; Ann Elaine Clarke; Marvin J Fritzler; Katherine A Buhler; Murray Urowitz; John Hanly; Yvan St Pierre; Caroline Gordon; Sang-Cheol  Bae; Juanita Romero-Diaz; Jorge Sanchez-Guerro; Sasha Bernatsky; Daniel Wallace; David Isenberg; Anisur Rahman; Joan T Merrill; Paul R Fortin; Dafna Gladman; Ian N Bruce; Michelle Petri; Ellen Ginzler; Mary Anne Dooley; Rosalind Ramsey-Goldman; Susan Manzi; Andreas Jonsen; Graciela Alarcon; Ronald van Vollenhoven; Cynthia Aranow; Meggan Mackay; Guillermo Ruiz-Irastorza; Sam Lim; Murat Inanc; Kenneth C Kalunian; Soren Jacobsen; Christine A Peschken; Diane L Kamen; Anca Askanase; Jill Buyon; David Sontag; Karen H Costenbader

doi:10.1136/ard-2022-223808

Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes

May Yee Choi^*
, Irene Chen
, Ann Elaine Clarke
, Marvin J Fritzler
, Katherine A Buhler
, Murray Urowitz
, John Hanly
, Yvan St Pierre
, Caroline Gordon
, Sang-Cheol Bae
, Juanita Romero-Diaz
, Jorge Sanchez-Guerro
, Sasha Bernatsky
, Daniel Wallace
, David Isenberg
, Anisur Rahman
, Joan T Merrill
, Paul R Fortin
, Dafna Gladman
, Ian N Bruce

Michelle Petri, Ellen Ginzler, Mary Anne Dooley, Rosalind Ramsey-Goldman, Susan Manzi, Andreas Jonsen, Graciela Alarcon, Ronald van Vollenhoven, Cynthia Aranow, Meggan Mackay, Guillermo Ruiz-Irastorza, Sam Lim, Murat Inanc, Kenneth C Kalunian, Soren Jacobsen, Christine A Peschken, Diane L Kamen, Anca Askanase, Jill Buyon, David Sontag, Karen H Costenbader

^*Corresponding author for this work

Inflammation and Ageing

Research output: Contribution to journal › Article › peer-review

387 Downloads (Pure)

Abstract

OBJECTIVES: A novel longitudinal clustering technique was applied to comprehensive autoantibody data from a large, well-characterized, multinational inception systemic lupus erythematosus (SLE) cohort to determine profiles predictive of clinical outcomes.

METHODS: Demographic, clinical, and serological data from 805 SLE patients obtained within 15 months of diagnosis and at three- and five-year follow-up were included. For each visit, sera were assessed for 29 ANA immunofluorescence patterns and 20 autoantibodies. K-means clustering on principal component analysis-transformed longitudinal autoantibody profiles identified discrete phenotypic clusters. One-way ANOVA compared cluster enrolment demographics and clinical outcomes at ten-year follow-up. Cox proportional hazards model estimated the hazards ratio (HR) for survival adjusting for age of disease onset.

RESULTS: Cluster 1 (n=137, high frequency of anti-Sm, anti-U1RNP, AC-5 (large nuclear speckled pattern), and high ANA titres) had the highest cumulative disease activity and immunosuppressants/biologics use at year ten. Cluster 2 (n=376, low anti-dsDNA and ANA titres) had the lowest disease activity, frequency of lupus nephritis, and immunosuppressants/biologics use. Cluster 3 (n=80, highest frequency of all five antiphospholipid antibodies) had the highest frequency of seizures and hypocomplementemia. Cluster 4 (n=212) also had high disease activity and was characterized by multiple autoantibody reactivity including to anti-histone, -dsDNA, -ribosomal P, -SSA/Ro60, -SSB/La, -Ro52/TRIM21, -PCNA, and -centromere B). Clusters 1 (adjusted HR 2.60 [95%CI: 1.12-6.05], p=0.03) and 3 (adjusted HR 2.87 [95%CI: 1.22-6.74], p=0.02) had lower survival compared to Cluster 2.

CONCLUSION: Four discrete SLE patient longitudinal autoantibody clusters were predictive of long-term disease activity, organ involvement, treatment requirements, and mortality risk.

Original language	English
Journal	Annals of the Rheumatic Diseases
Early online date	21 Apr 2023
DOIs	https://doi.org/10.1136/ard-2022-223808
Publication status	E-pub ahead of print - 21 Apr 2023

Access to Document

10.1136/ard-2022-223808

ChoiM2023Machine
This article has been accepted for publication in Annals of the Rheumatic Diseases, 2023 following peer review, and the Version of Record can be accessed online at: http://dx.doi.org/10.1136/ard-2022-223808. © Author(s) (or their employer(s)) 2023
Accepted author manuscript, 988 KBLicence: Creative Commons: Attribution-NonCommercial (CC BY-NC)

Cite this

Choi, M. Y., Chen, I., Clarke, A. E., Fritzler, M. J., Buhler, K. A., Urowitz, M., Hanly, J., St Pierre, Y., Gordon, C., Bae, S.-C., Romero-Diaz, J., Sanchez-Guerro, J., Bernatsky, S., Wallace, D., Isenberg, D., Rahman, A., Merrill, J. T., Fortin, P. R., Gladman, D., ... Costenbader, K. H. (2023). Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes. Annals of the Rheumatic Diseases. Advance online publication. https://doi.org/10.1136/ard-2022-223808

@article{1b0f4ca44c224dd4bec7cf6a047163d7,

title = "Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes",

abstract = "OBJECTIVES: A novel longitudinal clustering technique was applied to comprehensive autoantibody data from a large, well-characterized, multinational inception systemic lupus erythematosus (SLE) cohort to determine profiles predictive of clinical outcomes. METHODS: Demographic, clinical, and serological data from 805 SLE patients obtained within 15 months of diagnosis and at three- and five-year follow-up were included. For each visit, sera were assessed for 29 ANA immunofluorescence patterns and 20 autoantibodies. K-means clustering on principal component analysis-transformed longitudinal autoantibody profiles identified discrete phenotypic clusters. One-way ANOVA compared cluster enrolment demographics and clinical outcomes at ten-year follow-up. Cox proportional hazards model estimated the hazards ratio (HR) for survival adjusting for age of disease onset. RESULTS: Cluster 1 (n=137, high frequency of anti-Sm, anti-U1RNP, AC-5 (large nuclear speckled pattern), and high ANA titres) had the highest cumulative disease activity and immunosuppressants/biologics use at year ten. Cluster 2 (n=376, low anti-dsDNA and ANA titres) had the lowest disease activity, frequency of lupus nephritis, and immunosuppressants/biologics use. Cluster 3 (n=80, highest frequency of all five antiphospholipid antibodies) had the highest frequency of seizures and hypocomplementemia. Cluster 4 (n=212) also had high disease activity and was characterized by multiple autoantibody reactivity including to anti-histone, -dsDNA, -ribosomal P, -SSA/Ro60, -SSB/La, -Ro52/TRIM21, -PCNA, and -centromere B). Clusters 1 (adjusted HR 2.60 [95\%CI: 1.12-6.05], p=0.03) and 3 (adjusted HR 2.87 [95\%CI: 1.22-6.74], p=0.02) had lower survival compared to Cluster 2. CONCLUSION: Four discrete SLE patient longitudinal autoantibody clusters were predictive of long-term disease activity, organ involvement, treatment requirements, and mortality risk. ",

author = "Choi, \{May Yee\} and Irene Chen and Clarke, \{Ann Elaine\} and Fritzler, \{Marvin J\} and Buhler, \{Katherine A\} and Murray Urowitz and John Hanly and \{St Pierre\}, Yvan and Caroline Gordon and Sang-Cheol Bae and Juanita Romero-Diaz and Jorge Sanchez-Guerro and Sasha Bernatsky and Daniel Wallace and David Isenberg and Anisur Rahman and Merrill, \{Joan T\} and Fortin, \{Paul R\} and Dafna Gladman and Bruce, \{Ian N\} and Michelle Petri and Ellen Ginzler and Dooley, \{Mary Anne\} and Rosalind Ramsey-Goldman and Susan Manzi and Andreas Jonsen and Graciela Alarcon and \{van Vollenhoven\}, Ronald and Cynthia Aranow and Meggan Mackay and Guillermo Ruiz-Irastorza and Sam Lim and Murat Inanc and Kalunian, \{Kenneth C\} and Soren Jacobsen and Peschken, \{Christine A\} and Kamen, \{Diane L\} and Anca Askanase and Jill Buyon and David Sontag and Costenbader, \{Karen H\}",

year = "2023",

month = apr,

day = "21",

doi = "10.1136/ard-2022-223808",

language = "English",

journal = "Annals of the Rheumatic Diseases",

issn = "0003-4967",

publisher = "BMJ Publishing Group",

}

Choi, MY, Chen, I, Clarke, AE, Fritzler, MJ, Buhler, KA, Urowitz, M, Hanly, J, St Pierre, Y, Gordon, C, Bae, S-C, Romero-Diaz, J, Sanchez-Guerro, J, Bernatsky, S, Wallace, D, Isenberg, D, Rahman, A, Merrill, JT, Fortin, PR, Gladman, D, Bruce, IN, Petri, M, Ginzler, E, Dooley, MA, Ramsey-Goldman, R, Manzi, S, Jonsen, A, Alarcon, G, van Vollenhoven, R, Aranow, C, Mackay, M, Ruiz-Irastorza, G, Lim, S, Inanc, M, Kalunian, KC, Jacobsen, S, Peschken, CA, Kamen, DL, Askanase, A, Buyon, J, Sontag, D & Costenbader, KH 2023, 'Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes', Annals of the Rheumatic Diseases. https://doi.org/10.1136/ard-2022-223808

TY - JOUR

T1 - Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes

AU - Choi, May Yee

AU - Chen, Irene

AU - Clarke, Ann Elaine

AU - Fritzler, Marvin J

AU - Buhler, Katherine A

AU - Urowitz, Murray

AU - Hanly, John

AU - St Pierre, Yvan

AU - Gordon, Caroline

AU - Bae, Sang-Cheol

AU - Romero-Diaz, Juanita

AU - Sanchez-Guerro, Jorge

AU - Bernatsky, Sasha

AU - Wallace, Daniel

AU - Isenberg, David

AU - Rahman, Anisur

AU - Merrill, Joan T

AU - Fortin, Paul R

AU - Gladman, Dafna

AU - Bruce, Ian N

AU - Petri, Michelle

AU - Ginzler, Ellen

AU - Dooley, Mary Anne

AU - Ramsey-Goldman, Rosalind

AU - Manzi, Susan

AU - Jonsen, Andreas

AU - Alarcon, Graciela

AU - van Vollenhoven, Ronald

AU - Aranow, Cynthia

AU - Mackay, Meggan

AU - Ruiz-Irastorza, Guillermo

AU - Lim, Sam

AU - Inanc, Murat

AU - Kalunian, Kenneth C

AU - Jacobsen, Soren

AU - Peschken, Christine A

AU - Kamen, Diane L

AU - Askanase, Anca

AU - Buyon, Jill

AU - Sontag, David

AU - Costenbader, Karen H

PY - 2023/4/21

Y1 - 2023/4/21

N2 - OBJECTIVES: A novel longitudinal clustering technique was applied to comprehensive autoantibody data from a large, well-characterized, multinational inception systemic lupus erythematosus (SLE) cohort to determine profiles predictive of clinical outcomes. METHODS: Demographic, clinical, and serological data from 805 SLE patients obtained within 15 months of diagnosis and at three- and five-year follow-up were included. For each visit, sera were assessed for 29 ANA immunofluorescence patterns and 20 autoantibodies. K-means clustering on principal component analysis-transformed longitudinal autoantibody profiles identified discrete phenotypic clusters. One-way ANOVA compared cluster enrolment demographics and clinical outcomes at ten-year follow-up. Cox proportional hazards model estimated the hazards ratio (HR) for survival adjusting for age of disease onset. RESULTS: Cluster 1 (n=137, high frequency of anti-Sm, anti-U1RNP, AC-5 (large nuclear speckled pattern), and high ANA titres) had the highest cumulative disease activity and immunosuppressants/biologics use at year ten. Cluster 2 (n=376, low anti-dsDNA and ANA titres) had the lowest disease activity, frequency of lupus nephritis, and immunosuppressants/biologics use. Cluster 3 (n=80, highest frequency of all five antiphospholipid antibodies) had the highest frequency of seizures and hypocomplementemia. Cluster 4 (n=212) also had high disease activity and was characterized by multiple autoantibody reactivity including to anti-histone, -dsDNA, -ribosomal P, -SSA/Ro60, -SSB/La, -Ro52/TRIM21, -PCNA, and -centromere B). Clusters 1 (adjusted HR 2.60 [95%CI: 1.12-6.05], p=0.03) and 3 (adjusted HR 2.87 [95%CI: 1.22-6.74], p=0.02) had lower survival compared to Cluster 2. CONCLUSION: Four discrete SLE patient longitudinal autoantibody clusters were predictive of long-term disease activity, organ involvement, treatment requirements, and mortality risk.

AB - OBJECTIVES: A novel longitudinal clustering technique was applied to comprehensive autoantibody data from a large, well-characterized, multinational inception systemic lupus erythematosus (SLE) cohort to determine profiles predictive of clinical outcomes. METHODS: Demographic, clinical, and serological data from 805 SLE patients obtained within 15 months of diagnosis and at three- and five-year follow-up were included. For each visit, sera were assessed for 29 ANA immunofluorescence patterns and 20 autoantibodies. K-means clustering on principal component analysis-transformed longitudinal autoantibody profiles identified discrete phenotypic clusters. One-way ANOVA compared cluster enrolment demographics and clinical outcomes at ten-year follow-up. Cox proportional hazards model estimated the hazards ratio (HR) for survival adjusting for age of disease onset. RESULTS: Cluster 1 (n=137, high frequency of anti-Sm, anti-U1RNP, AC-5 (large nuclear speckled pattern), and high ANA titres) had the highest cumulative disease activity and immunosuppressants/biologics use at year ten. Cluster 2 (n=376, low anti-dsDNA and ANA titres) had the lowest disease activity, frequency of lupus nephritis, and immunosuppressants/biologics use. Cluster 3 (n=80, highest frequency of all five antiphospholipid antibodies) had the highest frequency of seizures and hypocomplementemia. Cluster 4 (n=212) also had high disease activity and was characterized by multiple autoantibody reactivity including to anti-histone, -dsDNA, -ribosomal P, -SSA/Ro60, -SSB/La, -Ro52/TRIM21, -PCNA, and -centromere B). Clusters 1 (adjusted HR 2.60 [95%CI: 1.12-6.05], p=0.03) and 3 (adjusted HR 2.87 [95%CI: 1.22-6.74], p=0.02) had lower survival compared to Cluster 2. CONCLUSION: Four discrete SLE patient longitudinal autoantibody clusters were predictive of long-term disease activity, organ involvement, treatment requirements, and mortality risk.

U2 - 10.1136/ard-2022-223808

DO - 10.1136/ard-2022-223808

M3 - Article

SN - 0003-4967

JO - Annals of the Rheumatic Diseases

JF - Annals of the Rheumatic Diseases

ER -

Machine learning identifies clusters of longitudinal autoantibody profiles predictive of systemic lupus erythematosus disease outcomes

Abstract

Access to Document

Fingerprint

Cite this