Seed and soil hypothesis of metastasis (Paget)

The claim. The distribution of metastases is not random and is not determined by anatomy alone. Disseminated cancer cells reach many organs but survive and grow only in those whose microenvironment provides the right growth factors, matrix, vessels and immune tolerance; the seed and the soil must both be right. Later extensions add that the primary tumour prepares the soil in advance (the pre-metastatic niche) and that seeds can lie dormant in unsuitable soil for years.

Who and when. Paget, The distribution of secondary growths in cancer of the breast, Lancet 1889, from an autopsy series of 735 women. James Ewing argued in 1928 that blood flow and lymphatic anatomy explained the pattern instead. Isaiah Fidler revived Paget's idea in the 1970s by selecting melanoma lines that metastasised preferentially to lung, and reviewed the hypothesis in 2003 and, with Langley, in 2011. Kaplan and Lyden described the pre-metastatic niche in 2005; Minn and Massagué identified genes that direct breast cancer to lung in the same year.

Evidence for. Organ tropism is reproducible: prostate cancer to bone, uveal melanoma to liver, lung cancer and melanoma to brain, colorectal cancer to liver and lung. Selected tumour variants home to specific organs in animals, and gene signatures that predict organ-specific metastasis exist. Bone marrow-derived cells and tumour-secreted factors and exosomes prepare distant niches before tumour cells arrive. Disseminated tumour cells are found in bone marrow of patients who never develop bone metastases, so arrival is not enough. Adjuvant bisphosphonates, which alter the bone soil, reduce bone recurrence and breast cancer death in postmenopausal women in the 2015 meta-analysis of randomised trials.

Evidence against and limits. Ewing was partly right: the liver is the first capillary bed for colorectal cancer via the portal vein and the lung is the first for most other cancers, and mechanical arrest explains a share of the pattern. The molecular definition of a receptive soil is still incomplete, and predicting which patient will develop which metastasis is not yet possible.

Predictions that held or failed. Held: metastasis is inefficient and organ-selective; dormancy in an unsuitable soil is real; changing the soil can prevent metastasis in bone. Failed or unfulfilled: drugs that block niche formation in other organs have not reached the clinic; denosumab did not improve disease-free survival in early breast cancer despite its effect on bone.

Therapies that came from it. Adjuvant zoledronic acid and other bisphosphonates in postmenopausal early breast cancer, bone-targeted agents in prostate cancer, and the surveillance strategies that follow each cancer's known destinations. It is the metastatic application of the microenvironment view and links to clonal evolution through the selection of metastasis-competent clones.

Status: established. A hypothesis that waited eighty years for its evidence and is now the organising principle of metastasis biology, with Ewing's purely mechanical account superseded but retained as a contributing factor.