Ch.10 — The Question of Risk • 201 
analysis cietermines a probability \ aluo tor eacb 
sli’[)— e.^.. in ti^uro :Ui slop II tbn [)i'()bability ot 
es('a|)f can be estimated based on tbe bistorical 
record of experiments with micro- organisms. 
Depeiuling on tbe degi'ee ot containment, tbe 
j)i'obabilit\ \aries. It is almost certain that 
expc'riments on an op('n beneb top, using no 
precautions, will result in some escape* to tbe 
surrouiuling en\ ironment— a much less likely 
e\ent in maximum containment facilities. (See 
table 3.1.) 
Two points sbould bi* noted, first, eacb prob- 
ability can be minimized by appropriate control 
measures. Second, tbe probability that tbe final 
e\ ent w ill occur is etiual to oi' less likeb than tbe 
least likely link in tbe chain, bec'ause tbe |)rob- 
al)ilities must be multiplied together, if the 
probabiliu of an\ single step is zero, tbe prob- 
ability of the final outcome is zei'o: the chain of 
e\ ents is broken. 
THE STATl S OF THE CI HKENT .\SSES.SME,\T 
OF PHYSICAL RISK 
.\ successful I'isk assessment sbould pro\ ide 
information about tbe likelihood and magnitude 
of damage that might occur under gi\en cir- 
cumstances. It is clear that tbe more types of 
damage that are identified, the moi'e risk assess- 
ments must be carried out. 
Figure 36.— Flow Chart to Establish Probability of 
Harm Caused by the Escape of a Micro-Organism 
Carrying Recombinant DNA 
Event 
Probability 
I. Inadvertent incorporation of hazardous gene 
into micro-organism 
II. Escape of micro-organism into environment 
f 
III. Multiplication of micro-organism and 
establishment in ecological niche 
4 
IV. Infection of man 
f 
V. Production of factor to cause disease 
P 
1 
^2 
P 
3 
P 
4 
P 
5 
NOTE: Ps will always be smaller than any of the other probabilities. 
SOURCE; Office of Technology Assessment. 
.Although the original charter of RAC under- 
.scored the importance of a risk assessment pro- 
gram, it was not until f979 that the details of a 
formal program were published. For 5 years, 
risks were assessed on a case-bv-case basis 
through: 1) experiments carried out under con- 
tract from iMIH, 2) experiments that were de- 
signed for other purposes but which proved to 
be reUnant to tbe c|uestion of risk, and 3) con- 
ferences at which findings were examined. 
From tbe start, it was difficult to design ex- 
periments that could supply meaningful infor- 
mation— e.g., bow does one test tbe possibility 
that "massive ecological disruptions might 
occur?” Or that a new bacterium with harmful 
unforseen characteristics will emerge? Still 
some experiments were proposed. But because 
tiiese exfjeriments bad to be approximations of 
tbe actual situation, tbe applicability of their 
findings was debated. Here too, experts could 
and did disagree— not about tbe findings them- 
seb es, but about their interpretation. 
For exani|)le, in an important experiment de- 
signed to test a "worst case situation,” a tumor 
\ irus called polyoma w as found to cause no 
tumors in test animals when incorporated into 
E. coli.^* Since just a few molecules of the viral 
DNA are know n to cause tumors when injected 
directly into animals, it was concluded that 
tumor \iruses are noninfectious to animals 
when incorporated into E. coli. If polyoma virus, 
which is the most infecti\ e tumor virus known 
for hamsters, cannot cause tumors in the rDNA 
state in E. coli, it is unlikely that other tumor 
\ iruses w ill do so. This conclusion has had wide- 
spread, but not unanimous, acceptance. It has 
been argued that there might be "something 
special” about polyoma that prevents it from 
causing tumors in this altered state; other 
tumor viruses might still be able to do so. At one 
meeting of RAC, in fact, it was suggested that 
experiments with several other viruses be car- 
ried out to confirm the generality of the finding. 
But how many more viruses? What is enough? 
=M. A. Israel, H. VV. Chan, W. P. Rowe, and M. A. Martin, "Molec- 
ular Cloning of Polyoma V'irus DNA in Escherichia Coli: Plasmid 
V'ector Systems," Science 203:883-887, 1979. 
'Some combinations of free plasmid and tumor virus DNA did 
cause infections. 
