Accepted: 10 August 2017 DOI: 10.1111/cyt.12484
ORIGINAL ARTICLE
Presence of cancer cells in gastric lavage of gastric cancer patients as an indicator of advanced disease, predictor of tumour aggressive phenotype and independent prognostic factor for poor survival: The endoluminal metastatic pathway of gastric cancer and GL0/GL1 classification E. Virgilio1
| E. Giarnieri2 | M. R. Giovagnoli2 | M. Montagnini2 | A. Proietti2 |
R. D’Urso2 | G. Nigri1 | P. Mercantini1 | G. Ramacciato1 | M. Cavallini1 | G. Balducci1 1
Department of Medical and Surgical Sciences and Translational Medicine, Faculty of Medicine and Psychology, University “Sapienza”, St. Andrea Hospital, Rome, Italy 2 Department of Clinical and Molecular Medicine, Faculty of Medicine and Psychology, University “Sapienza”, St. Andrea Hospital, Rome, Italy
Correspondence Edoardo Virgilio, Department of Medical and Surgical Sciences and Translational Medicine, Faculty of Medicine and Psychology, University “Sapienza”, St. Andrea Hospital, Rome, Italy. Emails:
[email protected]; edoardo.
[email protected]
Objective: As of 2017, the pathobiology of gastric cancer (GC) is far from fully understood; consequently, new methods of basic and advanced research have been proposed and tested. The presence (GL1) vs absence (GL0) of malignant cells exfoliated in gastric lavage (GL) of GC patients was formerly evaluated with diagnostic intent but not for staging or prognostic assessment. We investigated this hitherto unreported application of cytopathology. Methods: GL was preoperatively and prospectively collected from 80 GC patients and cytologically analysed. The results were compared with the classic clinicopathological features of GC and related to survival. The prognostic value of GL1 was assessed through univariate and multivariate analyses. Results: GL1 was detected in 36 samples (45%) and correlated with advanced tumour depth (T3-T4), lymphatic metastasis (N+), distant metastasis (M1) and lymphovascular invasion (LVI1; P=.0317, .0024, .003 and .0028, respectively). Overall survival (OS) was significantly shorter for GL1 (23 months) vs GL0 patients (42 months; P=.005) and GL1 vs GL0 T1 subjects (12.6 vs 47.8 months, P=.0029). Univariate analysis revealed that GL1, N+, M1, LVI1 and advanced stage were significantly associated with OS. Multivariate analysis assessed GL1 as the only independent prognostic factor for worse OS and progression-free survival (P=.0013 and .0107). Conclusions: In the present study, GL1 was correlated with advanced disease, aggressive tumour behaviour and poor prognosis. Although additional studies are needed to confirm these findings, the GL0/GL1 classification can be applied to GC patients to achieve higher accuracy in staging, prognostic stratification and treatment selection. KEYWORDS
gastric adenocarcinoma, gastric cancer, gastric cytopathology, gastric lavage, gastric washing
1 | INTRODUCTION
most frequently diagnosed cancer worldwide (after lung, breast, prostate and colorectal cancer) and third leading cause of tumour-
Although it has been almost 2 decades since the decrease in inci-
related death following lung and liver carcinoma.1 Different from
dence of this disease, gastric cancer (GC) still represents the fifth
Japan and Korea, where patients are routinely screened, typically
Cytopathology. 2018;29:41–48.
wileyonlinelibrary.com/journal/cyt
© 2017 John Wiley & Sons Ltd
|
41
42
|
VIRGILIO
diagnosed with early GC (EGC), subsequently treated using an endoscopic approach and destined to a favourable prognosis, in the rest of the world, this tumour is still detected at late stages and associ1,2
ET AL.
2.3 | Cytopathological analysis of GL GL samples were centrifuged twice for 10 minutes at 1500 g; for
In addi-
initial fixation, haemolytic and mucolytic agents were previously
tion to the lack of screening, unfavourable outcomes are likely to
added to samples that were particularly rich in blood or mucus. Fol-
reflect the limited current knowledge of the pathogenesis and patho-
lowing permanent fixation with Bouin solution, the same technician
biology of this disease as well as the high degree of clinical hetero-
(M.M.) stained the smears according to the Papanicolaou method,
geneity of patients suffering from the same stage of disease.
followed
Consequently, researchers have been investigating GC at different
cytopathologists (E.G. and M.R.G.).10 Cytological diagnoses were ren-
levels using basic clinical and molecular oncology studies, and among
dered using the conventional diagnostic nomenclature for nongynae-
these studies, amelioration of the present staging systems represents
cological cytological specimens and classified into five groups;
ated with high mortality (723 000 global deaths in 2012).
2-5
by
a
microscopic
evaluation
by
two
experienced
Currently, there is a long list of
normal exfoliated cells; slight changes in normal morphology; altered
new interesting prognostic factors pending validation.6-8 In the pre-
cells without pathognomonic features of malignancy; atypical cells;
sent study, we elected to use gastric lavage (GL) of GC patients as
and cells with pathognomonic features of malignancy and subse-
the starting point for scientific assessment. Considering the scarcity
quently arranged in clusters.11 The cytological findings were inter-
of pertinent literature, we verified whether cytopathological analysis
preted referring to the formerly known bioptic diagnosis of gastric
of GL could lead to scientific progress in terms of clinicopathological,
adenocarcinoma and expressed in positive or negative terms: if the
staging and prognostic knowledge and therapeutic implication.
cytological features were determined as malignant with histological
a common and notable effort.
characteristics, the cytological diagnosis was considered conclusive
2 | MATERIALS AND METHODS 2.1 | Nasogastric intubation
for GC without any need for additional studies. Three samples considered suspicious for malignancy were further investigated for diagnostic clarification through immunohistochemistry using the antibody anti-CEA (clone: 11-7; Dako, Glostrup, Denmark); among these sam-
Between April 2012 and September 2016, 80 patients with a preop-
ples, only one case showed antibody recognition and was elucidated
erative histology of gastric adenocarcinoma were recruited for this
as clearly neoplastic. The following morphological criteria were con-
prospective observational study in compliance with the principles
sidered characteristic of malignancy: nuclear changes (atypia, aniso-
embodied in the Declaration of Helsinki. Informed consent was
cariosis); increased and/or abnormal mitotic figures; high nucleus-to-
obtained prior to enrolment; 38 subjects were included in a previous
cytoplasm ratio; nucleolar hypertrophy or multiplicity; highly con-
9
study on GC. All participants were fitted with a nasogastric (NG)
densed nuclear chromatin; cytosolic vacuoles (signet-ring cells); pleo-
tube during their hospital stay at the Division of General and Emer-
morphism; hypertrophy; presence of aggregates; and pseudopapillary
gency Surgery of St Andrea Hospital, Faculty of Medicine and Psy-
(Figure 1).10,11 All GCs were histologically evaluated according to the
chology, “Sapienza” University, Rome, Italy. In 71 cases, the NG
7th AJCC TNM Staging System.12
device was inserted at the beginning of intervention prior to tumour manipulation, as palliative care in eight cases and prior to endoscopic submucosal dissection (ESD) of EGC in one case.
2.4 | Statistical analysis Statistical analysis was performed using MedCalc for Windows, ver-
2.2 | GL: the procedure
sion 16.8.4 (MedCalc Statistical Software, Ostend, Belgium). Different distributions of continuous variables among groups were
The stomach cavity was flushed through the NG pipe with 500 mL
calculated using Student t test, whereas discrete variables were com-
of a saline solution: considering its distensible nature, we reasoned
pared using Pearson’s chi-square test. For survival, we used the
that this cut-off value was sufficient to bathe the inner surface of
Kaplan-Meier method for calculation and graphical representation,
the entire viscus and improve the feasibility of collecting cancer cells,
log-rank test for comparison between patients with positive and (one- and two-way) for univariate analy-
irrespective of the tumour gastric site. Obstructing cardial carcino-
negative GL cytology,
mas (Siewert type I, II and III) and gastric stump cancers represented
sis, and Cox proportional hazards model for multiple regression anal-
sole exceptions to this general rule; in such cases, we adopted a
ysis. Four types of survival were evaluated: median overall (or
minimum quantity of 250 mL. Asepsis was maintained during the
observed) survival rate (OS), representing the time from the day of
entire procedure using sterile gloves and containers to minimise fun-
sampling to death from any cause; progression-free survival (PFS),
gal and bacterial contamination, which could interfere with subse-
elapsed from sampling until metastatic (that is not recurrent) pro-
quent cytological examination. All GLs were performed and
gression or death from any cause; time to tumour progression (TTP),
immediately transported (within a few minutes) to the Cytology
which differed from PFS by considering only cancer-related deaths;
Department of our hospital by the same operator (E.V.) to avoid
and disease-free (DFS) or recurrence-free (RFS) survival, representing
sampling errors and time-dependent alterations.
the time elapsed from sampling to any recurrence, metastasis or
ANOVA
VIRGILIO
|
ET AL.
(A)
(B)
(C)
(D)
(E)
(F)
43
F I G U R E 1 (A, B) Clusters of malignant cells shed from a well differentiated gastric adenocarcinoma (Papanicolaou stain, 9100 oil immersion). (C) Gastric adenocarcinoma showing mucin-secreting cells with eccentric nuclei (Papanicolaou stain, 9100 oil immersion). (D, E) Adenocarcinoma of the stomach with signet ring cells (Papanicolaou stain, 9100 oil immersion). (F) Clusters of malignant cells shed from a well differentiated gastric adenocarcinoma (Papanicolaou stain, 9100 oil immersion)
death from any cause. P values ≤.05 were considered statistically sig-
carcinomas (G4). According to the Lauren classification, there were 51
nificant.
intestinal (63.75%), 26 diffuse (32.5%) and three mixed (3.75%) types of GC. A total of 20 cancers (25%) had signet-ring cells, and among the
3 | RESULTS 3.1 | Clinicopathological features
64 evaluable cases, angiolymphatic and perineural invasion were observed in 32 (50%) and 19 (29.7%) cases, respectively. Tumour depth was observed in 13 cases with T1 (EGC; 16.25%), 12 cases with T2 (15%), 26 cases with T3 (32.5%) and 29 cases with T4 (36.25%). A
The patient population included 45 men (56.25%) and 35 women
total of 60 patients had lymph node metastasis (N+ including N1, N2
(43.75%). The median age was 69.5 years (range: 38-91). Cancer was
and N3; 75%), and 18 patients had distant metastasis (M1; 22.5%).
antropyloric in 39 cases (48.75%), cardial in 23 cases (28.75%) and cor-
The metastatic lymph node ratio (MLR) was 0 in 20 patients (31.25%),
pofundic in 18 cases (22.5%); among the latter, eight lesions were gas-
MLR1 (ie, >0-0.3) in 19 patients (29.68%), MLR2 (>0.3-0.6) in 17
tric stump carcinomas. Altogether, 72 patients were submitted to
patients (26.56%) and MLR3 (>0.6) in eight patients (12.5%). The
surgery: 64 patients underwent open surgery with curative intent and
pathological stage was stage 1A and 2B in 11 patients (13.75%), 1B
D 1.5 lymph node dissection (30 total gastrectomies, 30 distal gastrec-
and 3C in eight patients (13.75%), 2A in three patients (3.75%), 3A in
tomies with Roux-en-Y reconstruction, 4 Billroth II partial gastrec-
seven patients (8.75%), 3B in 14 patients (17.5%) and 4 in 18 patients
tomies), seven patients with inoperable GC were managed using
(22.5%). For the 17 patients (21.25%) who received neoadjuvant treat-
palliative gastroenterostomy (two cases) and nutritional Witzel
ment, chemotherapy was administered in 14 patients (EOX, ECF and
jejunostomy (five cases), and one patient was subjected to neoadju-
DCF were the frequently adopted combinations used herein) and com-
vant treatment following exploratory laparoscopy. At histology, the
bined radio-chemotherapy was administered in three patients. For the
surgical margins showed infiltration in seven cases (11%). A total of
52 patients administered adjuvant treatment, radio-chemotherapy was
eight patients were non-surgical; among these individuals, seven sub-
employed in 23 cases, chemotherapy alone was employed in 17 cases
jects with late GC were provided with palliative care through a naso-
and irradiation alone was employed in 12 cases.
duodenal tube (six cases) or oesophageal stent (one case), whereas one EGC patient underwent ESD at endoscopy. As for cancer grade, eight patients had well-differentiated carcinomas (G1), 10 patients had
3.2 | Cytopathology
moderately differentiated carcinomas (G2), 52 patients had poorly dif-
The cytological examination of GL demonstrated the presence and
ferentiated carcinomas (G3) and 10 patients had undifferentiated
absence of cancer cells (GL1 vs GL0) in 36 (45%) and 44 (55%)
44
|
VIRGILIO
ET AL.
patients, respectively. Notably, GL1 correlated with the parameters
48.4 months, P=.0004), PFS (16.6 vs 41.7 months, P=.0033) and
of tumour depth (T3-T4 vs T1-T2), lymph node metastasis (N+ vs
DFS (14.6 vs 41.7 months, P=.0013) but not TTP (24.8 vs
N0), distant metastasis (M1 vs M0) and lymphovascular invasion
50.8 months, P=.1812). Similarly, compared to Stage 1 GL0 patients,
(LVI1 vs LVI0) with statistical significance (respectively P=.0317,
Stage 1 GL1 participants showed shorter OS (15.5 vs 46.7 months,
.0024, .003 and .0028; Table 1). Notably, as for lymph node metas-
P=.0004), PFS (22.3 vs 48.4 months, P=.0054), TTP (23.3 vs
tasis classification, GL1 patients showed a significant relative risk of
55 months, P=.0169) and DFS (13.4 vs 48.4 months, P
