| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
a Department of Cardio-Vascular and Thoracic Surgery, U.C.L. Saint Luc Hospital, Brussels, Belgium
b Department of Pathology, U.C.L. Saint Luc Hospital, Brussels, Belgium
c Department of Pulmonary Medicine, U.C.L. Saint Luc Hospital, Brussels, Belgium
Received 5 September 2007; received in revised form 21 December 2007; accepted 16 January 2008.
* Corresponding author. Address: Cardio-Vascular and Thoracic Surgery Unit, U.C.L. Saint Luc Hospital, Université catholique de Louvain, Avenue Hippocrate 10, B-1200 Brussels, Belgium. Tel.: +32 2 7646107; fax: +32 2 7648960. (Email: Poncelet{at}chir.ucl.ac.be).
 |
Abstract |
|---|
 Top Abstract 1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
Key Words: Lung cancer • Statistics • Survival analysis
|
1. Introduction |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
Several prognostic factors for recurrence have been identified.
Among others factors that characterized tumor behavior, histological features of tumor ‘aggressiveness’ (vessel invasion, permeation) have been widely investigated across all stages. Tumor vascular invasion can be venous, arterial or lymphatic, while nervous invasion is most often localized to the perineural area. The prognostic value of tumor permeation has already been demonstrated both in breast [3], colo-rectal [4] and head and neck cancers [5].
Though sometime contradictory in lung cancer studies, most authors have accepted a positive correlation between those histological features and survival [6–10].
For the very early stage patients, if proven accurate, such postoperative pathological substaging could help in identifying groups of patients most at risk for recurrence or metastasis and therefore might guide clinicians in using adjuvant chemotherapy.
The purpose of this single-center study was to assess the prevalence of tumor permeation at large in a selected subgroup of completely resected patients with pathologically proven stage I or II NSCLC, and then to assess its prognostic significance with respect to cancer recurrence and patient survival.
|
2. Material and methods |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
Resection was classified R0 (macro- and microscopically complete) in 334 patients (97%), R1 (microscopically incomplete) in 12 patients (2.1%), including five positive bronchial margins. No patient was classified R2 (macroscopically incomplete) [12]. We selected for this survival analysis only the 334 patients who had a complete (R0) resection. There were 55 females and 279 males included in the study. Their clinico-pathological data are summarized in Table 1 . For this study, to avoid a potential confounding effect of adjuvant chemotherapy, data inclusion was closed in 2004 since it is only after the publication of the IALT trial results [13] that adjuvant chemotherapy became standard practice in our group for all suitable stage Ib and II patients.
|
2.2 Surgical strategies
The surgical procedures consisted of 237 lobectomies (70.9%), 18 bilobectomies (5.4%), and 70 pneumonectomies (21%). Segmentectomy was performed in one patient and wedge resection was performed in eight patients. Indications for lesser resection (2.7%) were limited pulmonary reserve in all cases.
Complete homolateral mediastinal lymph node dissection was performed in all cases and node stations were labelled according to the American Thoracic Society guidelines [14]. There were 4430 lymph nodes available for pathological examination with a mean number of 14.1 resected lymph nodes per patient (SD 9.9).
2.3 Pathologic evaluation
Tumor was evaluated in all 334 patients according to the World Health Organization 1999 classification for NSCLC both for histology and grade by one of the authors (B.W.) [15]. After localization and size measuring, the specimens were initially planed with a cryostat, serially sectioned (3–4 mm) and embedded, followed by standard hematoxylin and eosin staining. All resected lymph nodes were either formalin fixed (later period) or Bouin fixed (earlier period). The sections were than analyzed with special attention given to intra-tumoral evidence of tumor invasion to vascular vessels, lymphatics, and neural structures. Tumor vascular invasion (artery, vein or lymphatic) was defined as tumor cells infiltrating the vessels directly or neoplastic thrombosis of the vessel. The presence of neural invasion was defined as tumoral involvement of the epineurium at least.
2.4 Selection of predictor variables
Predictor variables studied for their potential impact on prognosis were age, sex, the histology of the tumor, the grade of differentiation, T status, N status, disease's pathological stage, the presence or absence of vessel and/or perineural permeation and the type of resection.
2.5 Data collection and patient follow-up
All the information necessary for the study was collected from the patient operative reports, the hospitalization charts and our thoracic surgery database. Follow-up was completed from 09/2006 until 05/2007 and was done through contacts with the referring pulmonary physician, the primary care physician or patient's family when appropriate. Survival data were cross-checked with the national registry database. As for survival, follow-up was 100% complete until closure of the study. However, a precise cause of death was only available in 119/185 reported deaths and only those patients were analyzed in term of cancer-free survival.
2.6 Statistical analysis
Survival was calculated from the date of the first surgery to the date of follow-up (or death). For patients presenting recurrences, the number of days was calculated from the date of pulmonary resection to the first documentation of either locoregional or distant recurrence. For cancer-free survival, patients still alive free from cancer and patients who died from cancer-unrelated causes were censored at the time of either follow-up or death.
For analysis of descriptive statistics and categorical variables, Chi-square or Fisher's exact test were used as appropriate whereas for analysis of continuous variables, Student's t-test was used. Survival analysis was performed according to Kaplan–Meier method. Univariate analysis by the log rank test and Cox regression model were used to compare survival, identify predictors of death and calculate hazard ratio. The level of statistical significance was set at a p value < 0.05. All statistical analyses were performed with SPSS version 14.0 software (SPSS, Inc., Chicago, IL)
|
3. Results |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
As described in Section 2, only completely resected patients were selected for prognostic factor analysis.
Histology revealed squamous cell carcinoma in 168 patients (50.3%), adenocarcinoma in 152 patients (45.4%) and undifferentiated carcinoma in 14 patients (4.3%). According to the grade of differentiation, there were 115 well differentiated (34.4%), 162 moderately (48.5%) and 57 poorly (17.1%) differentiated tumors.
3.2 Intra-tumoral permeation
Overall, 48 patients (14.4%) were diagnosed to have intra-tumoral vessel or perineural permeation and 286 patients had none. The most frequent type of permeation found was vascular permeation (n
= 35), followed by lymphatic permeation (n
= 10) and perineural (n
= 9). A few patients had more than one type of intra-tumoral permeation. We tried to identify clinico-pathological factors that positively correlated to the presence of intra-tumoral permeation. As shown in Table 2
, neither the mean age nor the mean tumor diameter of the group of patients with permeation significantly differed from the 286 patients without permeation. No significant correlation was found between tumor permeation and sex, or between node-positive and node-negative patients.
|
Other statistically significant positive correlations were found for tumor pathological stage (p = 0.02) and for the grade of differentiation (p = 0.03). Indeed, the higher the stage or the lower the tumor differentiation, the higher the prevalence of intra-tumoral permeation.
3.3 Predictors of overall survival (Table 3)
At a median follow-up time of 69 months, a total of 149 patients were still alive, whereas 185 patients had died.
|
The 5-year actuarial survival for female and male were 61.6% and 50.6%, respectively. Node-negative and node-positive (as stage I and II) patients had a 5-year survival of 56.2% and 39.9%. Finally, patients who underwent a lobectomy had a better 5-year survival than those with pneumonectomy (58.3% vs 31.5%).
As shown in Fig. 1 , the 3- and 5-year overall survival were 64.5% and 53.7% for those patients without permeation, and 60.3% and 44% for patients with tumor permeation. The median survival time for the former (72.1 months) was nearly twice as long as the one for patient with permeation (42.3 months). However, by Cox univariate analysis, tumor permeation was not found to be a significant factor of survival (HR 1.11, 95% CI [0.74–1.66]).
|
|
As shown in Fig. 2 , tumor permeation was again not a significant factor of cancer-free survival, with 3- and 5-year disease-free survival of 81.4% and 77.9% for those without permeation, and 66.1% at 3- and 5-year for patients with tumor permeation. The median cancer-free survival time was 46 months for patients without permeation and 30 months for those with permeation. Analysis of the subgroup with stage I disease (n = 208) revealed a median cancer-free survival time of 48 and 31 months, respectively (p = 0.29).
|
|
4. Discussion |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
Thus the literature review reveals that a wide variation in prevalence still persists and a clear explanation for those discrepancies is still lacking, though subjective histological assessment and the absence of objective qualitative and quantitative methods can be hypothesized.
Few studies have focused on clinico-pathological factors that correlate with the presence of vessel's tumor permeation. In our study, we found that permeation was positively correlated to the T factor, the grade of tumor differentiation as well as the stage. Those findings are in agreement with both the Pechet et al. [18] and Mineo and al. studies where both tumor diameter and differentiation were correlated with permeation [19].
In the present study, which is one of the largest reporting on stage I patients, we could not demonstrate a statistically significant prognostic impact of intra-tumoral permeation either on overall survival or on cancer-free survival. However, a 10% overall 5-year survival between patients with and without tumor permeation could well be seen as clinically relevant, even though statistical significance was not reached in this moderately-sized study. With respect to cancer-free survival, we should raise a word of caution in interpreting our results as a detailed cause of death was only available for 2/3 of the patients.
There are at least two hypotheses for this negative study. Firstly, our prevalence of permeation was mainly attributed to vascular permeation and several reports have emphasized the lack of prognostic significance in NSCLC studies [6,9]. Secondly, this retrospective study spans over more than 20 years, and even though we selected only patients with complete resection and systematic mediastinal lymph node sampling, the fact that PET FDG scanning was only introduced in the second half of the study period might have precluded us from detecting unforeseen metastatic disease which could have masked the potential prognostic impact of tumor permeation.
On the other hand, studies that only focused on either vascular or lymphatic permeation might have failed to clearly ascertain which of those components might have an undisputable impact on survival.
Finally, current positive studies still present with several weaknesses: in the one from Tsuchiya et al. [16], no data is given on completeness resection rate, the type of parenchymal resection is not analyzed and no information is given on cancer-free survival. In the study from Saijo et al. [20], stages I– III were all included and in their multivariate analysis, the pN factor was by far the most powerful predictor of death in the overall survival and recurrence-free survival analysis. Moreover, in the stage I subgroup analysis, the authors compared the only nine patients with extra-tumoral lymphatic tumor permeation. No information was given for stage I patients when lymphatic permeation (extra- and intra-tumoral) was analyzed.
In conclusion, in this large institutional study of early stage NSCLC, the presence of intra-tumoral permeation was correlated both to T, grade of differentiation, as well as to stage. However, in contrast to recent reports, we did not found that intra-tumoral permeation adversely affected long-term survival.
Further large-scale studies with recent patient cohorts and refined histological techniques will hopefully shed some light on this debated topic.
|
Appendix A |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
Dr R. Rami-Porta (Barcelona, Spain): You have concluded that whereas intra-tumoral permeation is associated to larger tumors, high undifferentiated grade and high pathologic status, it is not, by itself, prognostic of survival.
However, there are really striking differences in survival when we compare overall survival and also cancer-free survival. In the overall survival, there is a 16% 5-year survival difference between those patients with tumors with those pathologic features and those patients without the pathologic features that you have studied, and the difference in median survival is 30 months. When we look at cancer-free survival, the difference in 5-year survival is 12% and the difference in median survival is 16 months.
The curves that I’ve seen in the paper, both for overall survival and cancer-free survival, separate very nicely. The graph that you showed on overall survival is slightly different from the graph I’ve seen in the text. In the one you showed, both curves seemed to get together at the very end of the follow-up. It's not so in the graph that's in the paper. But anyway, it seems that the difference is consistent along all follow-up and is more evident in the cancer-free survival graph.
Why these differences are not statistically significant, I don’t know. You mentioned the scarce number of patients with these pathologic features. And although, you had a large series of patients compared with the series that you have mentioned in the discussion of your paper, when we start breaking down, the figures go smaller, and you only had 48 cases with these pathologic features.
And also when we analyze a single risk factor or a single potential risk factor, I think we may find this sort of discrepancy between your series and others reported. I imagine, for example, that maybe you have a cluster of patients with tumors that have a certain negative prognostic factor that has not been determined or it is still unknown, or maybe those tumors without pathologic features that you have studied have a higher SUV uptake as you showed this morning. Anyway, there may be other factors that have not been determined in the series.
In any case, I think that because these differences may be clinically relevant, even if there is not enough ground to recommend adjuvant therapy, perhaps, at least, these patients should be closely followed up after surgery just to see if you can pick up early recurrences or not.
As the authors concluded, I think that we still need larger series, maybe multi-institutional and prospective, with standardized pathologic study, to clarify the impact of these pathologic features in early stage non-small cell lung cancer.
I only have one question. You have not mentioned, and I have not seen it in the paper, whether these patients received adjuvant therapy after surgery.
Dr Poncelet: To answer easily to the last question, none of those patients had adjuvant chemotherapy since we only started adjuvant chemotherapy in 2004 in our university.
The main point I made from this study and working with our data, we have obviously one pathologist who is devoted to the lung, and so that person, who is the last author, I think really at each time has really made its best in diagnosing permeations. But when I saw our prevalence of 14.4%, I really felt that we may just increase the threshold of sensitivity by adding new detection techniques, such as elastin staining, for example. Because even though she's the only one to have concluded on the permeation, I think that we are lacking patients here and probably the group would be differently separated if we had done better immunohistochemistry studies. That would be my conclusion, I think.
|
Appendix B |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
|
Footnotes |
|---|
Presented at the 21st Annual Meeting of the European Association for Cardio-thoracic Surgery, Geneva, Switzerland, September 16–19, 2007. |
References |
|---|
|
TopAbstract1. Introduction2. Material and methods3. Results4. DiscussionAppendix AAppendix BReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
