Contiguous pattern spreading in patients with Hodgkin's disease.

Background: In 1966, Rosenberg and Kaplan hypothesized that Hodgkin's disease (HD) arises at a discrete primary site and subsequently spreads in a predictable manner via functionally contiguous lymph nodes. However, their results were not statistically evident. It was our aim to describe the spreading in the lymphatic system more precisely and to confirm their postulate.

Methods: Between 1971 and 1992, 297 patients underwent pathological staging for HD. Our subsequent evaluation was restricted to the 236 cases with cervical involvement (65 bilateral, 80 dextral and 91 sinistral), those with lymph nodes on the right side (65 + 80 = 145) being analyzed separately from those with tumours on the left (65 + 91 = 156). Spreading via the lymphatic system was assessed by scoring of the number of involved and uninvolved nodes in six regions, which are functionally contiguous in the lymph system but not necessarily anatomically neighboured. The number of 'gaps' (i.e. missed nodes) observed according to a systematic spreading model was compared with that expected (probability model) if a random course had been followed.

Results: Of the 156 patients with left cervical HD, 117 (75%) had para-aortic or spleen involvement, 90 (58%) had mediastinal involvement, 65 (42%) had right cervical involvement, 50 (32%) had axillary involvement and 23 (15%) had inguinal involvement. Of the 145 patients with right cervical HD, 112 (77%) had mediastinal involvement, 89 (61%) had para-aortic or spleen involvement, 65 (44%) had left cervical involvement, 44 (30%) had axillary involvement and 16 (11%) had inguinal involvement. In patients with left or right cervical lymph nodes, the proportions observed with gaps in the spreading were 37 and 27% (SE 7%), respectively, whereas the corresponding values of gaps expected in a probability model if a random course of spreading had been followed would have been 84 and 73% (P = 0.0001 and 0.0001, respectively).

Conclusion: Our data support the concept that HD spreads in a predictable manner via functionally contiguous lymph nodes. In patients with right cervical lymph nodes, HD spreads via the upper mediastinum and pulmonary hila to the upper abdominal nodes and the spleen. In those with left cervical tumours, HD spreads directly to the abdomen (bypassing the mediastinum), then upward again via the pulmonary hila and upper mediastinum to the neck region (bilateral involvement) and from here it proceeds to the axillary nodes. Finally the inguinal nodes are involved.