Arthritis & Rheumatism, Volume 62,
November 2010 Abstract Supplement

Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Atlanta, Georgia November 6-11, 2010.

An Algorithm Using Genome-Wide SNP Analysis for Prediction of Interstitial Pneumonia in RA Patients.

Nakamura3,  Takeshi, Koyano5,  Satoru, Funahashi5,  Keiko, Hagiwara3,  Takafumi, Miura3,  Takako, Okuda3,  Kosuke, Sagawa6,  Akira

Inoue Hospital, Takasaki, Japan
Izumihara Rheumatic and Medical Clinic, Kagoshima, Japan
Matsubara Mayflower Hospital, Kato, Japan
Matsuno Clinic for Rheumatic Diseases, Toyama, Japan
Research Institute of Joint Diseases, Kobe, Japan
Sagawa Akira Rheumatology Clinic, Sapporo, Japan
Shono Rheumatology Clinic, Fukuoka, Japan


Interstitial pneumonia (IP) is a serious complication for collagen diseases such as RA and is strongly associated with the prognosis of the disease. The presence of IP also limits the selection of medication. There is, however, no method for prediction of the risk of occurrence of IP. We established an algorithm based on genome-wide SNP analysis for prediction of IP (usual interstitial pneumonia: UIP; non-specific interstitial pneumonia: NSIP) in RA patients.

Patients and Methods:

The first population sample included 215 RA patients, the second included 115 patients: a total of 330 patients from 6 hospitals in different regions of Japan. Classification of IP was determined by three doctors (one physician and two radiologists) according to UIP and NSIP criteria. The first population included 14 UIP (average disease duration: 9.8 years), 27 NSIP (average disease duration: 11.7 years) and 174 non-IP (average disease duration: 10.7 years) patients, and the second included 10 UIP (average disease duration: 10.5 years), 16 NSIP (average disease duration: 11.5 years) and 89 non-IP (average disease duration: 11.0 years) patients. Genome-wide SNP genotyping was performed by HumanHap300K chip. Case-control analyses between 285,548 SNPs and classification of IP were examined by Fisher's exact tests. We selected 10 SNPs associated with IP, UIP, or NSIP, which were common in analyses of both the first and second population (p < 0.02). We then scored the relationship between each SNP and classification of IP, the estimated total score of 10 SNPs (estimated scoring in each SNP was as follows: homo allele in the majority in IP: +1 point, hetero allele: 0 point, and homo allele in the majority of non-IP: -1 point), and examined the relationships between IP and non-IP, and the total score.


Accuracy ((true positive+true negative)/total), specificity (true negative/(false positive+true negative)) and sensitivity (true positive/(true positive+false negative)) of the algorithm for UIP ranged from 86–91%. For NSIP, accuracy, specificity and sensitivity of the algorithm ranged from 87–93%. It is therefore suggested that the SNP algorithm can distinguish IP (UIP and NSIP) in individual RA patients.


This highly accurate algorithm using SNP analysis may be useful for the prediction of UIP and NSIP in individual patients, and, in this way, can contribute to establishing a strategy of treatment such as the selection of medication and prevention of IP in treated patients.

To cite this abstract, please use the following information:
Nakamura, Takeshi, Koyano, Satoru, Funahashi, Keiko, Hagiwara, Takafumi, Miura, Takako, Okuda, Kosuke, et al; An Algorithm Using Genome-Wide SNP Analysis for Prediction of Interstitial Pneumonia in RA Patients. [abstract]. Arthritis Rheum 2010;62 Suppl 10 :1739
DOI: 10.1002/art.29504

Abstract Supplement

Meeting Menu