Is there an epidemic of chronic lymphocytic leukaemia (CLL) in China?

Chronic lymphocytic leukaemia (CLL) is 10- to 20-fold less common in Asians (including Han Chinese) compared with persons of predominately European descent. Why is unknown but seems predominately genetic. We observed an increasing frequency of new cases of CLL at our Haematology Centre beginning 2011 and wondered why. Determine the cause(s) for this increased frequency. We interrogated the context of CLL diagnosis in 483 consecutive subjects seen at the Institute of Haematology of a large referral hospital in Beijing. 3 cohorts were considered based on why a CBC was done to establish the CLL diagnosis: (1) a CBC-testing situation unrelated to a health condition such as a routine annual health exam or application for employment or medical insurance (termed routine CBC); (2) an unrelated medical condition such as a cold, influenza, heart disease etc. (termed CBC for other disorders); and (3) signs and/or symptoms consistent with CLL such as lymph-adenopathy, hepato- or splenomegaly, fatigue, B-symptoms etc. (termed CBC for possible CLL). Data regarding context of CLL diagnosis were available for 389 subjects (81%). Proportions of subjects in the 3 cohorts were 44% (95% confidence interval [CI]; 39, 49%), 24% (20, 28%) and 32% (28, 37%). The proportion of subjects whose evaluation of CLL was prompted by an abnormal CBC not for possible CLL (cohorts 1 and 2) increased over the surveillance interval (r = 0.164; P = 0.001) as did median age at diagnosis (r = 0.207; P < 0.001). Age at diagnosis was correlated with probability of CLL being suspected because of an abnormal routine CBC (r = 0.249; P < 0.001); 42% (32, 53%) amongst subjects ≤50 years versus 86% (75, 92%; P < 0.001) among those >70 years. Consistent with this, older subjects were diagnosed at Rai stage-0 with asymptomatic disease compared with younger subjects (P < 0.001). Our data suggest much of the increased frequency of CLL at our centre and likely elsewhere in China predominately reflects ascertainment bias. Other variables may also operate.