Biomedical Engineering Reference
In-Depth Information
C
AVEATS
But, perhaps it is not so.
Caveats concerning models of dose-volume effects of tumors
To the extent that tumors are disorganized agglomerations of non-
communicating proliferating malignant cells, the assumptions in the
TCP model seem reasonable. However, the model makes other,
sometimes unstated, assumptions. First, that the malignant cells
within a given tumor do not communicate with one another, which the
bystander effect mentioned above puts into doubt. And then, that the
malignant cells within a given tumor are all equally radiosensitive. It
is clear that this is generally not the case. Different parts of the tumor
may have quite different oxygen tensions, and oxygenation alters cell
sensitivity markedly. In some tumors, the periphery may be well
oxygenated as compared with the interior of the tumor and, as the
minimum dose often tends to be near the tumor periphery, there is
probably a correlation between dose delivered and radiosensitivity.
The EUD concept has the attraction of great simplicity - but, it makes
the unlikely presumption that one can say everything important about
dose-volume effects with just one parameter.
The simple thought experiment presented in the section entitled “TCP
and minimum dose,” above, suggests pretty conclusively that the
conventional wisdom that predicts that TCP is determined by the
minimum dose delivered anywhere within the tumor cannot be
correct. Nevertheless, one cannot say that this point is proven. For
example, in Terahara
et al.
(1999), an attempt was made to correlate
local recurrence of skull base chordomas with measures of dose (such
as minimum and mean dose) and with EUD. Both minimum dose and
EUD (but not, surprisingly, mean dose) were, in a Cox multivariate
analysis, found to be good predictors of outcome! The reason for this
was clearly demonstrated: because rather uniform treatment
techniques were used, there was (as is often the case) a strong
correlation of several of the measures - in this case between
minimum tumor dose and EUD.
It seems precarious to rely on models of the dose-volume response of
tumors for anything but modest deviations from established
experience.
