The claim traces to a single 2010 paper. Its purpose was prognostic, not therapeutic. The authors aimed to find a metastasis-prone signature for early stage mismatch-repair proficient sporadic colorectal cancer patients for better prognosis and informed use of adjuvant chemotherapy, analyzing genome-wide expression profiles of 82 age-, ethnicity- and tissue-matched patients and healthy controls using the Affymetrix U133 Plus 2 array. The best classification model yielded a 54 gene-set (74 probe sets) with an estimated prediction accuracy of 71%.

The specificity, sensitivity, negative and positive predictive values of the signature are 0.88, 0.58, 0.84 and 0.65 respectively.

The GHK finding is one sentence at the end of that paper: querying the Connectivity Map with a subset (70%) of these genes shows that Gly-His-Lys and securinine could reverse the differential expressions of these genes significantly, suggesting that they have combinatorial therapeutic effect on the metastasis-prone patients.

Pickart's later review articles restate this as a suppression percentage: GHK suppressed RNA production in 70% of 54 human genes overexpressed in patients with an aggressive metastatic form of colon cancer and was active at a low nontoxic 1 micromolar concentration.

In 2010, Hong et al. used Broad Institute's Connectivity Map to find molecules that could inhibit metastatic colon cancer. The Connectivity Map contains expression profiles that were evaluated in five cancer cell lines, in response to 1,309 bioactive small molecules.

No source anywhere states 74%. The number 74 in this literature refers to probe sets, not a percentage.

METHODOLOGY AND CONTEXT

Study design: Retrospective microarray profiling of 82 matched patients and controls, followed by an in silico database query. This is a bioinformatics signature-matching exercise, not an experiment in which GHK-Cu was given to anything with cancer.

What the Connectivity Map actually is: a reference database of gene expression changes in cultured cancer cell lines exposed to compounds. The original CMap was built primarily in the breast cancer epithelial cell line MCF7, with a subset of perturbagens also profiled in prostate cancer line PC3 and the nonepithelial lines HL60 (leukemia) and SKMEL5 (melanoma). A cMap "hit" means a compound's expression fingerprint runs statistically opposite to a query signature. It does not measure tumor shrinkage, survival, or metastasis.

The compound tested: cMap profiled Gly-His-Lys, the bare tripeptide, in cell culture. The viral claim is about GHK-Cu, the copper complex sold in skincare and injectables. These are related but not identical, and no colon tissue or colon cancer patient was exposed to either in this work.

The 54 genes: these were selected by a machine learning classifier for predictive accuracy, not established as causal drivers. The authors describe them as node molecules including YWHAB, MAP3K5, LMNA, APP, GNAQ, F3, NFATC2, and TGM2, which integrate multiple bio-functions in various compartments into an intricate molecular network, suggesting that cell-wide perturbations are involved in metastasis transformation. A prognostic classifier gene is a marker. It is not the same as a driver.

Conflict of interest: the review literature amplifying this finding is dominated by one author with direct commercial interest. Pickart founded two companies to sell copper peptides, co-founding ProCyte, which he took to NASDAQ, and Skin Biology. He is also a named inventor on United States Patent 9,586,989, "Non-toxic skin cancer therapy with copper peptides."

Animal evidence: limited and old. Pickart's own review cites a mouse sarcoma model in which a mixture of GHK-copper 2+ and ascorbic acid produced a 60% reduction in tumor growth. That is a single non-colorectal rodent experiment, not replicated at scale.