The discovery of anaplastic lymphoma kinase (
ALK) gene
rearrangements in a small subset of lung adenoracimonas in
2007 led to the definition of a new molecular subgroup of non
small cell lung cancers (NSCLC) (
1). These
ALK-rearranged
tumors are most commonly adenocarcinomas, often with signet
ring mucinous cells, without mutation of EGFR or KRAS, and
preferentially arise in non- or light smokers, but some
ALK-rearranged
NSCLC cases do not fit this description. A number
of fusion variants have been identified; the most common being
EML4-ALK fusions which are formed from inversions within
the small arm of chromosome 2. The encoded proteins comprise
the N-terminal portion of EML4 and the intracellular catalytic
domain of ALK. A dimerization or oligomerization of these
chimeric proteins leads to a constitutive activation of the ALK
kinase domain (
2). Other described fusion partners of
ALK in
lung tumors comprise
KIF5B,
TFG and
KLC1 (
3-
5). Whether
the nature of the fusion partner has a biological significance
in terms of ALK subcellular localization and activation, and a
clinical significance in terms of response to treatment, is not
known at the present time.
Dramatic and prolonged responses have been obtained
in ALK-rearranged patients treated with the ALK and MET
inhibitor, crizotinib (PF-02341066), leading to the approval of
the drug by the FDA as the first licensed ALK inhibitor for ALKpositive
(ALK+) NSCLC.
Even if different methods can be used to detect ALK
rearrangements in lung tumors, the presence of more than 15%
of tumor cells with an ALK fluorescent in situ hybridization
(FISH)-positive pattern using break-apart probes remains the
reference used to prove the presence of an ALK rearrangement
in the patients enrolled in crizotinib studies.
In their study published in Cancer in January 2012 (
6),
Camidge and colleagues explore the correlations between the
percentage of
ALK+ cells and signal copy number as assessed by
FISH, and their association with response to ALK inhibition by
crizotinib.
In our routine ALK FISH practice, most of us (if not all) have
observed variability in the percentage of positive cells in ALK+
tumors, as well as the presence of a small percentage of positive
cells in non-rearranged tumors. For the authors, this cellular
heterogeneity has not a biological but a technical explanation,
and they give an astute demonstration of their hypothesis.
To start with, the authors point out that it is easier to detect
single red (3’) signals than a split pattern within a nucleus, and
that a rearrangement is more readily detected when the number
of signal copies is increased. Therefore, if cellular heterogeneity
is due to technical aspects only, the number of red signals (which
contribute to the single red and the split patterns of positivity)
should be positively correlated to the percentage of ALK+
cells in ALK+ tumors, fused signal copy number should not be
associated with the percentage of ALK+ cells, and there should
be a positive correlation between the number of green (5’)
signals (which contribute to both the split pattern of positivity
and the negative single green pattern) with ALK+ only in cases
with a split pattern of positivity.
Finally, to further rule out any biological explanation, there
should be no correlation between the percentage of ALK+ cells
present in ALK+ tumors and tumor shrinkage (as assessed by
Response Evaluation Criteria In Solid Tumors, RECIST) after
crizotinib therapy.
The authors therefore evaluated the percentage of ALK+ cells,
patterns of ALK positivity (split, single red or both), and signal
copy number (fused, isolated red and isolated green signals)
in relation with response to ALK inhibition in ninety ALK+
NSCLC patients. Out of these 90 patients, 30 received crizotinib
(phase 1 study) and had available response outcomes on therapy.
The main results show that the percentage of ALK+ cells
was comprised in an 18-100% range (mean =56%) in the 90
ALK+ patients, 24-94% (mean =59%) in the 30 crizotinibtreated
patients. Confirming the authors’ hypothesis, tumors
with a single red pattern of positivity, easier to detect, had a
significantly higher mean percentage of ALK+ cells than those
with a split pattern of positivity, even if the range of positive cells
was similarly wide. Increased copy number of isolated red signals
also strongly correlated with a higher percentage of ALK+ cells,
and isolated red signal copy number gain was stronger for the
single red than for the split pattern of positivity. The percentage
of positive cells was strongly correlated to isolated green signal
copy number only in tumors with a split pattern of positivity,
and fused signal copy number was negatively associated with
percentage of cells positive for a rearrangement.
On the biological side, mean maximal tumor shrinkage
as assessed per RECIST after Crizotinib treatment was 58%,
ranging from 0 to 100%. As expected by the authors, there was
no correlation between the percentage of ALK+ cells in ALK+
tumors and the maximal tumor shrinkage after treatment.
However, the number of patients was not sufficient to find any
significant difference between the extent of tumor shrinkage and
any pattern of positivity or copy number.
This article, together with the paper published by the same
team in Clinical Cancer Research in 2010 (
7), are two precious
paper for all who work in the
ALK FISH field, giving a certain
number of very valuable technical guidelines and explanations.
First of all, contrary to the theory of Martelli et al. for who
ALK rearrangements may represent a late oncogenic event,
resulting in the coproliferation of different ALK+ and ALK- negative
clones within established tumors, for Camidge et al.,
the negative cells within a tumor are rather false negatives, and in
the same way, the ALK+ cells in ALK negative tissues (tumor or
normal tissue) are false-positives. A certain number of technical
caveats can explain this phenomenon, but for the authors the
main explanation for false-negative signals is interobserver error,
especially in the cases of EML4-ALK rearrangements, which
lead to a separation of the two probes which can be missed.
However, the authors consider a separation of the two probes
to be positive if the splitting occurs by more than two signal
diameters. A splitting by more than one signal diameter seems
more appropriate in the case of ALK rearrangements, especially
for EML4-ALK, together with a comparison with another
technique; immunohistochemistry and/or RT-PCR, which are
very helpful in doubtful cases.
On the other hand, the authors rightly point out that false
positive patterns can be due to stretching of the DNA, leading to
an artificial separation of the two probes.
Taken together the data in this article show there is a need to
assess the response to treatment of the “ALK-positive tumors”
in order to define - or not - a positivity threshold above which
tumors respond to treatment. However, for the moment little is
known about the response to treatment of the various fusions
proteins; is the response to treatment the same for EML4-ALK
rearranged tumors than for tumors containing another fusion?
Do all the EML4-ALK variants respond in the same way to
therapy? And what about the response to various inhibitors? A
certain number of questions remain. The use of other techniques,
especially of RT-PCR or sequencing could allow a better
understanding of the biology of these various ALK fusions,
leading to another step in personalized medicine, and the advent
of high throughput sequencing methodologies will no doubt be
very helpful.