59 / 220
in other human malignancies
[26
e
28]
. Interestingly, we
found that patients whose management plan was highly
impacted by PET/CT had significantly worse PFS and
OS than the patients with low/no impact; in contrast,
patients with moderate impact did not differ from pa-
tients with no/low impact in terms of PFS or OS.
Notably, the survival curves of the patients with mod-
erate and no/low impact converged toward the end of
follow-up. This suggests that despite nodal upstaging,
PET/CT appropriately modified the treatment plan in
the patients with moderate impact. However, the pa-
tients whose management was highly impacted by PET/
CT failed to obtain a survival benefit because PET/CT
detected distant metastasis and SPCs. Detection of these
lesions in patients with HNSCC generally leads to
palliative treatment and death
[20]
.
The current clinical guidelines recommend perform-
ing PET/CT as an optional imaging modality in
advanced stage disease because of a low diagnostic yield
of PET/CT in patients with stage I or II disease
[17,29]
.
Indeed, we observed that PET/CT failed to detect the
lesion in 14.8% of patients with T1 tumours, due to
limitations of PET/CT in assessment of the early T1
stage
[30]
. However, we also found that PET/CT could
upstage several of our patients with CWU stage I
e
II
disease into stages III and IV; after these restaging,
significant differences in stage-related prognosis were
found, whereas the CWU stage I and II patients did not
differ in terms of prognosis. These PET/CT-induced
changes also changed the management of 9.8% of the
CWU stage I and II patients. The number of these pa-
tients is relatively small and results should be interpreted
carefully. Nevertheless, our findings suggest that PET/
CT may need to be implemented in the routine imaging
workup for initial staging in all patients with HNSCC,
not only those with advanced stage disease.
In this study, PET/CT staging was significantly more
accurate than CWU staging: it improved the staging
accuracy in 16.1% (40/248) of the patients. Our data are
similar with previous studies, which show that most
PET/CT-induced stage migration is the result of nodal
upstaging
[10,11,18]
. However, contrary to a previous
meta-analysis
[31]
, we did not find that CWU and PET/
CT differ significantly in terms of the accuracy with
which they detect SPCs. This may reflect the failure of
PET/CT to detect synchronous oesophageal cancer.
This was also observed by two recent studies
[29,32]
.
Thus, PET/CT may detect early oesophageal cancer in
HNSCC with particularly low sensitivity.
Our study had several limitations. First, the
CWU and PET/CT results were only validated in the
patients with discordant PET/CT and CWU stages and
in those in whom PET/CT changed the management
plan. Because changes of management were not ex-
pected in patients with identical PET/CT and CWU
stages, we did not evaluate these patients further. We
believe that this reflects the clinical practice of HNSCC
staging workup and does not alter our findings. Second,
our study cohort included heterogeneous tumour sites,
although they were histologically identical (HNSCC).
The disparities in T staging, natural courses, and clinical
behaviour in HNSCC patients may exist according to
the primary tumour location. Third, histopathological
confirmation was unavailable in some of the patients.
Fourth, this study was not designed to assess the cost-
effectiveness of PET/CT in initial routine imaging
workup. A randomised trial that addresses this issue is
warranted. Nevertheless, our results are valuable as they
Fig. 1. Kaplan
e
Meier curves of progression-free survival (PFS) (A) and overall survival (OS) (B) according to the effect of PET/CT-
induced changes in conventional workup (CWU) stage of the 248 eligible patients. Patients with upstaged diseases had worse PFS and
OS than those whose CWU stage did not change after PET/CT (3-year PFS
Z
56.8% versus 74.5%,
P
Z
0.043; 3-year OS
Z
61.3% versus
85.3%,
P
Z
0.006).
I.S. Ryu et al. / European Journal of Cancer 63 (2016) 88
e
96
39