- 2 - 
Attachment IV - Page 8 
agent were to prove hazardous, the risk would be borne by individuals 
other than the patient alone. This is a characteristic not shared 
by other therapeutic protocols, even those using chemical agents. 
Planned gene delivery systems, particularly those using retroviral 
vectors, may present a danger to the public health. Our concern 
is based on recent work on animal oncogenic retroviruses. Such 
viruses can transform normal cells into malignant cells of their 
hosts by a number of means including the introduction into those 
cells of a normal or slightly altered copy of a host gene. When 
this results in quantitative, qualitative, or temporally aberrant 
expression of the gene's protein product, a tumor can result. So 
far we have no way of predicting which host-like genes can act as 
oncogenes when presented to a cell out of normal physiological 
context, but the list currently includes genes specifying secreted 
growth factors, plasma membrane and cytoplasmic growth factor 
receptors, in addition to other common cellular proteins. It is 
presumed that animal retroviruses have picked up host genes in the 
course of infection during their evolutionary histories. If so, 
this must be a relatively rare event which would not necessarily 
render the recombinant virus oncogenic. Subsequent evolutionary 
events, such as an alteration of a single nucleotide in the viral 
version of the host gene leading to an amino acid change in the 
corresponding protein and possible oncogenicity, would occur more 
rapidly. It is well known that oncogenic retroviruses in animals 
can be transmitted "vertically" or "horizontally," that is between 
infected parents and offspring and from one individual to another 
within a population. 
Although no human retroviruses have yet been described that 
stimulate tumor formation by means of a host-like viral oncogene, 
there is no reason in principle why such agents could not arise 
naturally or by design. Perhaps the human species has been 
fortunate in harboring no retroviruses equipped to capture a 
genetic message from a host cell; however, genetic engineering 
could readily circumvent this situation and produce a retroviral- 
human gene recombinant. Indeed, this would be the desired goal 
in designing a somatic-cell gene therapy vector. Furthermore, 
such artificial recombinants might contain genes, like those for 
growth factors, that fall into both "therapeutic" and "oncogenic" 
categories. The failure of such recombinant retroviruses to 
exhibit oncogenicity or other types of pathogenicity in standard 
laboratory tests ("Points to Consider" I.B.2.C. (1) (c) ) is no 
guarantee that such recombinants are not a few short steps from 
becoming novel human pathogens. 
Although gene replacement therapy may ultimately prove to be an 
important tool in the management of certain serious diseases, there 
are few such diseases for which it is the only conceivable approach. 
Our concern is that existing technology applied to this end may 
raise more problems than it solves. We call on the Recombinant 
DNA Advisory Committee to convene a risk assessment conference 
on the use of retroviruses for vectors in somatic-cell gene therapy 
[ 35 ] 
