Volunteers' Recruitment and Questionnaires
In 2004 and 2005 the German Crohn's and Colitis Association (patient association with more than 20,000 members; DCCV e.V.), and the Competence Network Inflammatory Bowel Disease (medical expert network; Kompetenznetz Darmerkrankungen e.V.) published several calls for twin pairs using advertisements and the nationwide newsletter of the patient association. Individuals were asked to participate and complete a questionnaire if they had IBD and were born as 1 of a twin pair. Among others, questions were related to zygosity, medical history, social status, lifestyle (e.g., former and present smoking status), if they grew up together, and birth order.21 Special attention was paid to the chronic IBD subphenotype i.e., CD, UC, or other chronic inflammatory disease of the bowel.
Diagnosis of Crohn's Disease and Ulcerative Colitis
Diagnosis of IBD was confirmed by review of the patients' original medical records including ileocolonoscopies. Forty-three percent of the patients had at least 1 clinical visit including ileocolonoscopy performed at 1 of the investigators hospitals (A.R.), which ensured thorough diagnostics in these patients. Diagnosis of CD was confirmed when at least 2 of the criteria published by Landers et al22 were fulfilled. Diagnosis of UC was based on endoscopic appearance and continuity of inflammation, histology, and proven exclusive involvement of the colon (allowing backwash-ileitis). Patients were asked about age of onset of symptoms, age at time of diagnosis, location of disease, type of disease (e.g., stricturing, inflammatory), current treatment, and treatment since diagnosis. The nondiseased twins were evaluated for signs and symptoms of IBD and records of previous ileocolonoscopies were scrutinized.
Concordance describes the proportion of twins that share the same phenotype. The population casewise rate (i.e., “Proband concordance rate”) and the risk that a twin is affected if the co-twin is also affected were calculated according to the formula CPΓ = (2C1 + C2)/(2C1 + C2 + D) where C1 = number of concordant twin pairs in which the 2 affected members have been ascertained independently, C2 = number of concordant twin pairs in which only 1 twin is ascertained independently, D = number of pairs in which only 1 member is affected.23 The 95% exact confidence intervals for the concordance rates were calculated assuming a binomial distribution. By comparing the relative risks for the siblings of twins with CD and UC, we calculated whether genetic influence is of greater importance for 1 of the 2 diseases. According to the prevalence rates in the background population the relative risks were calculated by dividing the observed number of diseased co-twins with the expected numbers among the co-twins.24 To calculate the 95% confidence interval of the relative risk we assumed that the number of cases among the co-twins followed a Poisson distribution.19 Two relative risks were compared using the likelihood ratio test. Differences in weight and height between affected and nonaffected twins were calculated using the Wilcoxon signed rank test for zero median. By using McNemar's test for a case-control study with continuity correction, we evaluated whether smoking, birth order, and extraintestinal manifestations (EMs) were of significance for the discordance of IBD.
The univariate survival data from the Human Life-Table Database, Germany, for the period 2000–2002 was analyzed to estimate the number of twin pairs within the German population.25 The bivariate survival function in the correlated gammafrailty model follows the formula:
Here σ2 is the variance of frailty (nonobserved risk of mortality) and R is the correlation of the frailty for twins. The values of R and _2 were close to estimates obtained from the Danish twin study, σ2 = 1, RMZ = 0.5, RDZ = 0.25.26
The number of twin pairs with both twins survived to the age x is equal to NS(x,x), where N is the number of twin pairs at birth. The full number of twin pairs with both twins alive is equal to N ∑xS(x,x).
The population of Germany is now approximately 82,468,000 (inhabitants), the birth rate is 8.5%,27 each 80th birth is a birth of a twin pair, and 1/3 of twin pairs are monozygotic (MZ). That is, N &Ap; 82,468,000 × 0.0085 / 81 = 86,540 newborn twin pairs, with 28,847 MZ twin pairs and 57,693 dizygotic (DZ) twin pairs. Summing the number of twin pairs from 0 to 99 years old we were able to estimate a number of 207,891 MZ twin pairs and 413,321 DZ twin pairs in German population. Let P(D/H) = 1-P(H/H) be the conditional probability for diseased co-twin given healthy twin and P(D/D) = 1-P(H/D) the conditional probability for diseased co-twin given diseased twin. From the Bayes formula we have P(D/H) = P(H/D)*P(D)/P(H) &Ap; (1-P(D/D))P(D), where P(D) = 1-P(H) is the probability to find diseased individual in the population. In our case P(D) = 48/100,000 for CD and 80/100,000 for UC.
If we observe 207,891 MZ twin pairs, then in average we have 207,891*P(D) = 100 (166) CD (UC) diseased first MZ twins and 207,891 − 100 = 207,791 (207,891 − 166 = 207,725) CD (UC) healthy first MZ twins. Additionally, for diseased first MZ twins, we have 100*P(D/D) = 52 CD (166*P(D/D) = 46 UC) diseased co-twins and for healthy first MZ twins we have 207,791*P(D/H) = 47 CD (207,725*P(D/H) = 120 UC) diseased co-twins. Approximately we have 2*100 = 200 CD (2*166 = 332 UC) diseased individuals from MZ twins or 100*(2-P(D/D)) = 147 CD (166*(2-P(D/D)) = 286 UC) MZ twin pairs with at least 1 diseased twin.
If we observe 413,321 DZ twin pairs, then on average we have 413,321*P(D) = 198 CD (331 UC) diseased first DZ twins and 413,321 − 198 = 413,123 CD (413,321 − 331 = 412,990 UC) healthy first DZ twins. Moreover, for diseased first MZ twins we have 198*P(D/D) = 13 CD (331*P(D/D) = 10 UC) diseased co-twins and for healthy first DZ twins we have 413,123*P(D/H) = 185 CD (412990*P(D/H) = 320 UC) diseased DZ co-twins. Approximately we have 2*198 = 396 CD (2*331 = 662 UC) diseased individuals from DZ twins or 198*(2-P(D/D)) = 383 CD (331*(2-P(D/D)) = 651 UC) DZ twin pairs with at least 1 diseased twin. For P(D/D) values we used the proband concordance rates from Table 1.
Table 1. Mono- and Dizygotic Twin Pairs Concordant and Discordant for Crohn's Disease and Ulcerative Colitis
|Index Twin's Diagnosis||CD||UC||No IBD||Pair Concordance (%)||Proband Concordance (%)|
|Monozygotic|| || || || || |
| CD||11||0||20||35.5 (19.2–54.6)||52.4* (36.4–68.0)|
| UC||0||6||31||16.2 (6.2–32.0)||27.9 (15.3–43.7)|
|Dizygotic|| || || || || |
| CD||2||0||56||3.5 (0.4–11.9)||6.7 (1.9–16.2)|
| UC||0||1||62||1.6 (0.0–8.5)||3.1 (0.4–10.8)|
Differences resulting in a P-value of <0.05 were considered statistically significant.