| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Thoracic Surgery, University Hospitals Leuven, Leuven, Belgium
Received 31 July 2007; received in revised form 3 March 2008; accepted 4 March 2008.
* Corresponding author. Address: University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium. Tel.: +32 16 346820; fax: +32 16 346821. (Email: toni.lerut{at}uz.kuleuven.ac.be).
 |
Abstract |
|---|
 Top Abstract 1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
Key Words: Esophagectomy • Esophageal cancer • Elderly
|
1. Introduction and aim of the study |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
There is no established cut-off to define a patient as ‘elderly’ in relation to surgery, but most studies available so far set the age limit at 70 [5–14]: all these studies conclude that surgical resection is justified in selected aged patients, mainly those aged between 70 and 75 years. If esophagectomy for this group of patients is nowadays accepted, less is reported on patients over 75 [15–17]. This study focuses on the short- and long-term outcome after esophagectomy for cancer of the esophagus and gastro-esophageal junction in patients 76 years old and over. From the results an effort is made to perform a risk analysis for postoperative mortality by applying already acquired mathematical models for the prediction of mortality in surgery [18,19,20] and performing a multivariate analysis on collected data. Net survival and age-adjusted survival rates are calculated in order to estimate more precisely the life gain resulting from surgical treatment in this aged subset of patients.
|
2. Materials and methods |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
From January 1991 to May 2006, 108 consecutive patients over the age of 75 (that is 
76) with cancer of the esophagus or GEJ underwent surgical resection. Tumors of the hypopharynx and cervical esophagus as well as sub-cardia tumors were excluded from this analysis. Only patients for whom complete resection was deemed achievable and surgery had been performed with curative intent were included.
Preoperative work-up included: medical history and examination, oncological staging and functional evaluation. Staging investigations consisted of esophagogram (100% of the patients), endoscopy (100%), endo-ultrasound (92.5%), thorax and abdomen CT scan (96.2%), abdomen ultrasound (60.1%), PET or PET-CT (available since 1998 and 2002, respectively, performed on 66.6% of the patients). Further investigations (bronchoscopy, neck CT scan or ultrasound with node needle aspiration, bone scintigraphy, brain CT scan, mediastinoscopy, thoracoscopy, laparoscopy) were performed as indicated. Functional status was assessed by means of routine blood tests, electrocardiogram, and pulmonary function tests. Second line investigations (cardiac ultrasound, cardiopulmonary effort test, etc.) were performed when necessary, but more recently on a systematic basis.
The preoperative risk factors recorded included: anemia (hemoglobin less than 12 g/dl); weight loss more than 10% in the previous 3 months, history of tobacco or alcohol abuse, previous malignancies.
Comorbidities included: hypertension (defined as prescribed history of hypertension, systolic blood pressure higher than 140 and/or diastolic pressure higher than 90); cardiac dysfunction, defined as history of ischemic heart disease (need for medications, previous acute myocardial infarction, CABG and PTCA), unstable angina, cardiac failure, atrial fibrillation, pacemaker, previous heart transplant; pulmonary dysfunction, defined as FEV1 < 70% of predicted value and/or vital capacity (VC) < 80%; vascular dysfunction, defined as history of abdominal aorta aneurysm, vascular bypass, carotid stenosis and/or history of PTA, history of embolism; neurological dysfunction, defined as history of cerebrovascular accident (CVA) or transitory ischemic attack (TIA) or (history of) epilepsy or Parkinson; liver cirrhosis (pathologic diagnosis); chronic renal insufficiency or dialysis; diabetes mellitus on treatment.
Data regarding clinico-pathological aspects (tumor size, site, histology, grading, pTNM, completeness of resection, neoadjuvant therapy), surgical approach and extent of resection, postoperative morbidity and mortality, long-term outcome (raw and age-adjusted survival, disease specific survival, recurrence rate) of the patients were analyzed.
Cervical anastomoses were routinely monitored clinically, resuming oral diet on the fifth postoperative day. Upper GI series was performed only for patients with intrathoracic anastomosis (usually on the fifth postoperative day) or in any patient with suspected anastomotic fistula. Definition of anastomotic leak as a postoperative complication includes all those leakages at the anastomotic site that were detected either clinically or radiologically during the hospital stay.
Recurrent nerve palsy was detected on a clinical basis and confirmed by laryngoscopy during hospital stay.
Postoperative mortality was defined as any death occurring within 30 days of surgery or during the same hospitalization. Survival was calculated from the date of surgery.
The P-POSSUM [19] and O-POSSUM [20] scores were retrospectively applied to the study population as benchmarking for the mortality rate. The POSSUM [21] score (Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity), first described in 1991, was developed to allow risk-adjusted prediction of postoperative mortality in surgery, according to the physiological condition of the patients, severity of surgery, age and general health. It consists of a physiological part (evaluating 12 factors, 4 grades score) and an operative part (6 factors, 4 grades). The original logistic regression equation for mortality was found to overpredict the risk of postoperative death and was modified in 1998 by the Portsmouth group of Prytherch and co-workers [19], thus assuming the name of Portsmouth POSSUM (P-POSSUM) score, which showed to fit better the observed postoperative mortality. Also the O-POSSUM [20] (esophago-gastric POSSUM), specifically designed for upper gastrointestinal surgery in 2004 was retrospectively tested on our population.
Another score for prediction specifically of 30-day postoperative mortality, recently developed and published by Steyerberg et al. [18], was applied as well. In this model four variables are considered: age, comorbidity, neoadjuvant therapy and hospital volume. The sum of these factors as a score is entered into a formula to calculate the risk of 30-day mortality in terms of percentage (see Table 4).
|
Follow-up consisted of medical consultation every 3 months for the first year, thereafter every 6 months; serum markers and instrumental investigations (CT scan, PET scan, endo-US) once a year or on clinical indication. Follow-up data were recorded from the outpatient clinic medical notes and, when necessary, from enquiries obtained from both general practitioners/referring physicians and patients/patients’ families.
Multivariate analysis of pre-, intra- and postoperative data was attempted in order to find out factors possibly involved in determining postoperative morbidity and mortality (both by means of binary logistic regression) and long-term survival (by means of Cox regression). First, all the variables were tested univariately by means of Kaplan–Meier log-ranks (p < 0.05 were considered significant). In a second step, all the significant variables were stepwise forward entered (conditional entering at 0.05 level), thus producing the final model. As a last step, all the variables that were not withheld in the final model were entered in the model, one by one, and their effect on the model was checked. All calculations were performed using SPSS for Windows, version 10.
|
3. Results |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
The study population was composed of 76 (70.3%) males and 32 (29.6%) females. Mean age at the time of surgery was 79.5 (range 76–89); 38 patients (35.1%) were 80 years old or over.
About 80% (79.6%) of the patients presented at least one comorbid condition (Table 1 ), more commonly hypertension (42.6%), cardiac (38.9%) and pulmonary dysfunction (26.9%). Twenty percent had a history of previous malignancies.
|
|
Final pathological staging showed an equal distribution between early (0-I-IIa: 56 patients, 52%) and more advanced stages (II-b, III, IV: 52, 48%). Six (5.5%) patients underwent preoperative chemoradiotherapy, which obtained complete pathologic response in three cases, and down-staging in the remaining three.
Resection was complete in 90 patients out of 108 (83.3%), 13 had microscopic infiltration of margins of resection (R1) and five had macroscopic disease left behind (of these, one corresponding to primary tumor and four due to intraoperative finding of unforeseen organ metastases).
As a rule, patients were extubated in the operating theatre or within a few hours after the operation and all of them spent the first night in the recovery room, afterwards being transferred to the ward. Twenty-four patients (22%) were either admitted or re-admitted to the intensive care unit (ICU) because of major complications. For these patients, mean ICU stay was 26 days, median 17.5.
Postoperative morbidity occurred in 51.9% of the cases (Table 3 ) and was mostly represented by cardiopulmonary complications. Twenty-two percent of patients presented major complications, defined as requiring (re)-admittance to the intensive care unit. Surgical complications rate was 16.6%. The incidence of anastomotic leakage, the most threatening among surgical complications, was 2.8% (three patients). Re-operation for surgical complications was needed in three cases (2.7%): one for bleeding, one for chylothorax, one for tracheal tear due to the orotracheal tube.
|
Postoperative mortality did not decrease significantly when comparing the more recent period (1999–2006) to the former one (1991–1998): 7.4% versus 7.3%.
Mean hospital stay was 24 days (median 15.5, range 4–172).
When applied to our study population, the P-POSSUM [19] and O-POSSUM [20] predicted a postoperative mortality of 7.2% and 15.1%, respectively, thus the first fitted the actual mortality (7.4%), while the latter overestimated it by about two-fold.
When the 30-day mortality risk score developed by Steyerberg et al. [18] was applied to the study population, the predicted mortality was 6.8% (see Table 4 ), very close to the actuarial 5.5% 30-day mortality in our population. Indeed, six patients out of 108 (5.5%) died within 30 days from surgery, while two more died afterwards within the same hospitalization.
At last follow-up (November 2006) 65 patients were dead: eight due to postoperative mortality, 32 due to either local or metastatic recurrence, 13 due to causes unrelated to cancer, four from metachronous unrelated tumors, eight patients died from unknown cause. When calculating cancer specific survival, all the unknown causes of death were classified as cancer deaths. Forty-three patients are still alive: one with locoregional recurrence, one with metastatic recurrence, four with metachronous tumors, the rest without evidence of disease. The follow-up is complete for all patients (until death or until last follow-up). Mean follow-up is 28.5 months (range 0.1–149.5); if we consider only the patients who are still alive, mean follow-up is 37.2 months (range 6.5–149.5). Eighty patients (74%) have a follow-up of at least 3 years.
Overall and R0 five-year survival rates were 35.7% and 43.2%, respectively, while esophageal cancer specific R0 survival was as high as 51.7% (see Fig. 1A).
|
Multivariate analysis showed that cardiac dysfunction, R status and pathologic lymph nodal status were independent prognostic factors affecting long-term survival (Fig. 1B). The only factor independently related to postoperative major morbidity was ASA score 3, while factors related to postoperative mortality were age >81 years and two-field lymphadenectomy (Table 5
).
|
|
4. Discussion |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
Past reports [14] announced disappointing results for esophagectomy in the elderly, both because of encountered worse postoperative outcomes and perceived small survival benefit. More recent studies outlined acceptable results, with well selected aged patients showing neither significantly different postoperative mortality rate nor poorer long-term survival than younger counterparts after esophagectomy for cancer [5,6,8,10–13] (see Table 6 ). The vast majority of these studies set the cut-off to define a patient ‘elderly’ at the age of 70, the analyzed populations thus having a mean age around 73. Less has been published for patients over 75 or more [15–17]. Our aim was to investigate this ‘very old’ subset of patients, thus our cohort is composed of 76 years old and over patients, with a mean age of 78.9 and a number of octogenarians as high as 38 (35.1%).
|
Notable improvements both in clinical oncologic staging and in functional preoperative evaluation of patients have occurred in the last two decades, leading to a better selection of surgical candidates and therefore to possibly better outcomes. Various scores have been proposed and tested over the years in order to achieve a valid comprehensive risk stratification of patients preoperative conditions. We decided to test P-POSSUM [19] and O-POSSUM [20] scores with our population because of two reasons: first of all, the necessary data to calculate the scores were reliably available from our database; secondly these scores are also meant for surgical audit, so they conversely evaluate the quality of surgery as well.
When retrospectively applied to our population, P-POSSUM fitted very well the actual postoperative mortality rate, while the O-POSSUM score overestimated it by about two-fold. We found the 30-day mortality prediction score recently published by Steyerberg et al. [18] very practical and reliable. However, despite good overall calibration and discrimination on the whole cohort, all the risk stratification scores failed in predicting outcome on an individual basis. It has been recently suggested that, in order to achieve accurate individualized forecasts of operative risk in aged cancer patients, a single risk score is insufficient, and a sum of different scales and scores should be employed: the PACE (Preoperative Assessment of Cancer in the Elderly,) study [23], a prospective international study, with the aim at outlining the fitness of elderly surgical patients with malignancies, appeared to be able to define the correlation of some scores (ECOG Performance Status, Activities of Daily Living – ADL and Instrumental Activities of Daily Living – IADL) to short-term surgical outcome when applied to 215 elderly patients undergoing breast, colorectal, upper GI, and urogenital surgery for malignancy. However the number of patients receiving major surgery for gastroesophageal cancer was too small (nine patients) in this study to draw any meaningful conclusion. Unfortunately, most of the various tools for surgical risk assessment incorporated in the PACE must be tested preoperatively (that is in a prospective way), thus this new promising method could not be applied to our retrospective population.
The postoperative mortality of 7.8% places our series in the average of reported mortality (Table 6), which ranges from 1.9 to 10.9% for patients over 70. In a recent study [16] analyzing 751 patients (aged 50 to 79 vs 80) over a 10-year period, Moskovits and co-workers describe a mortality of 7.3% in the group 70–79 years old and 19.4% in the octogenarians group, thus concluding that age over 80 is related to higher mortality rate. If we consider our population, five deaths out of eight occurred in patients over 80, meaning a mortality rate of 4.2% in patients aged 76–80 and 13.1% in patients aged 80+. Others [15] have reported analogue or lower mortality rates [10]. However, it is debatable what can be considered ‘acceptable’ surgical mortality in front of an otherwise fatal malignancy such as esophageal cancer and so it is still controversial whether age over 80 per se is a bad prognostic factor for short-term outcome.
There is an apparent contradiction in literature reports: on one hand well-designed risk analysis studies accuse age of being an independent bad prognosticator for postoperative morbidity [24], mortality [24,25] and long-term survival [16], on the other hand other studies, particularly those with direct comparisons of older versus younger populations [5,6,8,10–13,17], refute these conclusions.
Some prospective [7,9] and retrospective [6–8] series focus on the recorded improvement over time in short-term outcome (i.e. morbidity and mortality) in aged patients undergoing esophagectomy, despite no significant increase in long-term survival. Improving in different aspects of management (case selection, better surgical techniques, minor blood loss, shorter operative time, ameliorated anesthetics and perioperative managements, decrease in fluid administration, peridural anesthesia with efficacious pain relief and better physiotherapy, aggressive use of bronchoscopy) reflect on the outcome of the whole population of surgically treated cancer patients, but even more evidently on those who are more fragile, those whose cardiopulmonary reserve is low and can show to be insufficient under stressful conditions.
We believe the load of evidence showing benefits from surgery for elderly patients is not negligible and further optimization in selection criteria and risk stratification will better clarify the advantage of aggressive treatment. This will require more prospective studies involving aged patients since only few have been published so far [7,9,23].
A common argument to justify suboptimal treatment in the elderly is a perceived small benefit in survival, considering short life-expectancy due to age-related illnesses. However, life-expectancy is on a constant growth: at present a 80-year-old man from Belgian Flanders is reasonably expected to live another 6.32 years; for a woman of the same age the life-expectancy would be 8.51 years (data collected from the Institute of Health of Belgian Flanders, available at http://www.wvc.vlaanderen.be/gezondheidsindicatoren/extra/uitleg/willems.pdf). In order to investigate how wide the life loss in our population was, we calculated the age-adjusted relative survival rates. The graph shown in Fig. 2 shows a visual comparison between the actual survival of our patients and the predicted survival of a general population matched for sex and age structure (that is, the survival our cohort could have reasonably expected if not affected with esophageal cancer). The life loss is represented by the gap between the actual and the predicted survival curves: so an overall 5-year survival of 35.7% (consistent with the rest of the literature, compare with Table 6), is actually much higher, because at that time about 40% of patients would have died anyhow. Interestingly, the two curves cross at about 7 years from surgery. This means two things: 1) the survival limiting factor in the first 7 years of follow-up is most likely esophageal cancer, not age; 2) those patients who were candidates for surgery and survived esophagectomy are naturally selected in the long term, because they are endowed with better cardiopulmonary function. A chance of cure should not be denied to these patients on the sole base of calendar age, but rather on the thorough assessment of biologic age.
|
We can deduce our rate of esophagectomy in the elderly from indirect clues. From the Flemish Cancer Registry Network (Flemish Cancer Registry Network 2006 booklet, available on line at http://www.tegenkanker.net/uploadedfiles/Kankerregistratie/Kankerincidentie/2000–2001.pdf, downloaded in May 2007) it appears that 24.5% of all new esophageal cancers in the period 2000–2001 occurred in patients aged 75, while this rate is as high as 45.9% if gastric cancer (including GEJ) is considered. These numbers, compared to the proportionally much lower rate of esophagectomies performed in the same age group of patients in the same period (8.9%), give an indirect idea of how large the population is of esophagus and GEJ elderly patients who are not referred to a specialist team.
Given the lack of adequate individual based operative risk stratification and the expected survival issues it becomes evident that the risk for postoperative mortality has to be weighed against the oncologic prognosis, i.e. the potential oncologic benefit of the surgical intervention. This study was not intended to assess the value of the clinical staging in the subset of patients aged over 75 years. But it is well known from the overall data that clinical staging despite improvements coming from EUS, PET, PET/CT is still suffering from a substantial lack of accuracy for both T and N staging, making the assessment of the oncologic prognosis still too unreliable.
As a result clinical expertise in both clinical staging and individual operative risk assessment as well as for the surgery itself are of paramount importance in the final decision making process as to the optimal therapeutic strategy. Such expertise is increasingly believed to be present in high volume centers displaying well established multidisciplinary expertise.
|
5. Conclusion |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
Still, a reliable individual risk analysis stratification to guide surgeons and oncologists in the decision making is missing: there is a significant lack in evidence regarding the treatment of cancer in the elderly population that is still under-enrolled in clinical trials [7]. The limit in the approach conducted so far is trying to apply to an older group the evidence that was provided by studies conducted on younger populations. The inclusion of aged patients in clinical trials for the treatment of esophageal cancer, or better, developing clinical trials especially designed for this subset of patients is fundamental in order to provide the scientific basis for the standardization of optimal multidisciplinary management of esophageal cancer in the elderly.
|
Appendix A |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
Dr S. Mattioli (Bologna, Italy): You have pointed out that the age is not the quality parameter because we usually select fit patients over 75. Some time when I am going to operate I ask myself if it is really correct for me to do extensive lymphadenectomies for example for this kind of patients, or could I do a quicker and simpler operation. Could you comment on that please?
Dr Internullo: It is a difficult balance, because the more limited operation you perform, maybe the higher rate of recurrence during follow-up, and we know that recurrence affects not only survival but also quality of life, which is fundamental in these patients. On the other hand you could argue that a more extensive procedure could carry too high a risk. In our population there was an equal distribution of early and more advanced stages: all of them had a lymphadenectomy half a two-fields, very few a three-field. But all of them were operated with the intention to have a radical resection through the same approach as we have for younger people. When you deal with an early stage, I think it's reasonable to perform a more limited lymphadenectomy.
Dr S. Cassivi (Rochester, MN): This was an excellent presentation. What you have demonstrated to us is that in very well selected patients, oesophagectomy can be done for the benefit of those elderly patients. Can you give us an idea of how you selected these patients here at the University of Leuven.
Dr Internullo: Unfortunately since I have taken these data from a surgical database in which only operated patients are included, and this is a retrospective analysis, we are missing the path that led to the decision to refer patients.
Dr Cassivi: I understand. But based on your experience here during your fellowship with Dr Lerut and his colleagues, can you give us an insight into the factors that you and they were taking into account when these elderly patients were considered for oesophagectomy?
Dr Internullo: I showed in my presentation the functional tests that were performed. I can tell you that for example people with a FEV1 lower than 70% were not excluded. People with cardiac comorbidities i.e. with ejection fraction lower than 60%, were not excluded. Patients were considered fit after studying carefully their cardiopulmonary function. The selection was tailored on the patient. I have no general criteria to show you, because the data regarding selection were not collected.
Dr R. Berrisford (Exeter, U.K.): Could I ask you a question about your methods? In your score is age a risk factor? So you use a score that includes age as a risk factor to determine if age isn’t. Another thing as I get older, old age seems further and further away. And certainly if we are operating on older and older people I used to think 79 was old, and now 84 is old, because we are seeing older and older people. Age is a continuous variable and you lose much of your strength of your data by de-commissioning that. So I call in to question your risk factor analyses but it would be a much more powerful paper if you analyse age in... and tell us were the evident differences arising in age rather than at 76 years. And that is a big problem with many papers, so your paper would be stronger if you maybe not risk adjust with a factor age and you look at age as a continuous variable.
Dr Internullo: Most of the papers on oesophagectomy in the elderly are based on populations with a mean age of 74–76. We wanted to push the cut-off a bit forward. Our population has a mean age of 80. It is one of the largest so far, we had 38 patients over the age of 80.
Dr Berrisford: It would be interesting to see the extremes. The problem with your risk adjustment is mortality of 50.3%. One question is the validity of your risk adjustment in 2007. It is a very high mortality. Your risk adjustment is maybe too high and it favours your conclusion.
Dr Internullo: I think that the O-POSSUM score is not reliable risk adjustment tool actually. The other two tests (the P-POSSUM score and the Steyerberg score) did very well, that is they fitted our actual mortality. These two latter models have been tested on very different age groups in the literature and they have already shown to be reliable. Our study didn’t aim at verifying the quality of the risk models we used.
Dr T.W. Rice (Cleveland, OH): The O-POSSUM as you know is based on historical series of patients in the UK, we have a mortality about 9% and that was not all patients; that was a standard mortality of 9 to 10%. You quoted that the overall mortality was 7.4% which is good for all the patients. Is the reason for mortality for patients over 75 today, is that acceptable for a hospital mortality today in patients over 75?
If it isn’t what do you do now to get a better selection in patients over 75, are there new tests?
Dr Internullo: The main problem is that we are missing a way to select correctly the surgical candidates. I think that if we had not operated on some of these patients, the mortality would be of course much lower. I think that the mistake in the approach we have used so far in the elderly population is that we are trying to apply to the elderly the evidence that was gained by studies performed on a younger population. They are not the same in elderly patients, even though they can show normal results in cardiopulmonary tests performed at baseline, that is what we are doing – the cardiopulmonary tests before surgery – often reveal a cardiopulmonary reserve shorter than expectable under stressful conditions that is after the operation.
Dr Rice: Do you use the LCL in your risk stratification now on these older patients?
Dr Internullo: I would like to make a prospective selection of different risk models. On this topic there was a nice abstract presented at the ASCO Meeting in 2006. It was a multicentric study called PACE (Preoperative Assessment in Cancer Elderly) it considered different risk stratification systems for surgical treatment of cancer in the elderly and it showed that most of the available models considered in singles, were not good enough to predict mortality. But this study dealt with kind of ‘minor’ surgery for cancer (i.e. breast, prostate, etc.). It should be tested on oesophageal cancer surgery.
|
Footnotes |
|---|
Presented at the 15th European Conference on General Thoracic Surgery, Leuven, Belgium, June 3–6, 2007. |
References |
|---|
|
TopAbstract1. Introduction and aim...2. Materials and methods3. Results4. Discussion5. ConclusionAppendix AReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
