| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Section of Thoracic Surgery, Division of Cardiothoracic Surgery, Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama
Accepted for publication January 7, 2008.
* Address correspondence to Dr Cerfolio, Section of Thoracic Surgery, Division of Cardiothoracic Surgery, University of Alabama at Birmingham, 703 19th St S, ZRB 739, Birmingham, AL (Email: robert.cerfolio{at}ccc.uab.edu).
Presented at the Fifty-fourth Annual Meeting of the Southern Thoracic Surgical Association, Bonita Springs, FL, Nov 7–10, 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Methods: A nested case-control study was conducted during 7 years using a prospective database of patients with non–small cell lung cancer. Younger patients (<45 years of age) were matched 1:2 with older patients for stage, sex, performance status, and type of resection.
Results: There were 762 patients (254 were <45 years old, 508 controls were older). The median time from initial symptom to thoracic surgical consultation was significantly longer for those younger than 45 years (6.5 versus 2.8 weeks; p < 0.001). Younger patients were more likely to be symptomatic at the time of diagnosis (89% versus 68%; p < 0.001) and less likely to be smokers (45% versus 78%; p < 0.001). Kaplan–Meier analysis showed the time between diagnosis and treatment, symptoms, maximum standardized uptake value on positron emission tomography, and smoking status impacted survival. Only symptoms and smoking status impacted survival on Cox proportional hazards survival analysis among completely resected patients; 5-year survival was lower in the younger group compared with the older group (51% versus 62%; p = 0.037).
Conclusions: Despite similar stages and tumor characteristics patients younger than 45 years of age with non–small cell lung cancer have a significantly worse prognosis than older patients. Although they are more likely to be symptomatic, younger patients have a greater delay in seeking thoracic surgical care. These data should be considered in the treatment strategy offered to younger patients with non–small cell lung cancer.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Matching
Young patients (defined as <45 years at the time of diagnosis) were matched 1:2 with older patients (defined as >45 years) for stage, sex, performance status, and type of pulmonary resection. Outcomes included hospital length of stay, major morbidity (defined in this manuscript as pneumonia, myocardial infarction, stroke, bleeding, adult respiratory distress, and the need for reintubation), and operative mortality. Time until presentation for treatment (defined as time from either initial symptom or from time of incidental finding on imaging for asymptomatic patients to the time the patient presented to the thoracic surgeon) and presence or absence of symptoms at the time of diagnosis were also recorded. Five-year survival was also compared.
Staging
Staging was performed as previously described at length [5]. In brief, patients were pathologically staged after having clinical staging using integrated PET/CT and CT scan. All suspicious N2, N3, or M1 locations (defined after PET as areas with a maximum standardized uptake value [maxSUV] >2.5) were biopsied before pulmonary resection. Mediastinoscopy was used to biopsy suspicious lymph nodes in the paratracheal area (stations 2R, 4R, 2L,4L, and top of 7), and endoscopic transesophageal ultrasound [6] was used to biopsy suspicious left-sided paratracheal lymph nodes (4L), subcarinal (7), periesophageal (8), and inferior pulmonary ligament nodes (9). The techniques used for endoscopic ultrasound fine-needle aspiration have also been previously described [7]. Pathologic review was performed using standard techniques, and immunohistochemical staining was performed when appropriate. A pathologic stage was generated using the T, N, and M classification system according to the international staging system [8].
Statistics
Analysis was conducted using SAS software 9.0 (SAS Corp, Cary, NC). Continuous data are presented as median values, and categorical data are presented as percentages. Fisher's exact test or Pearson's 
2 test was used to assess categorical data, and the Wilcoxon test was used to evaluate continuous variables. Survival estimates were derived by Kaplan–Meier analysis, and Mantel–Haenszel log-rank test was used to assess differences in survival between the young and elderly groups. Stratified log-rank analyses and Cox proportional-hazards modeling were used to investigate and adjust for major prognostic and stratification factors. A two-sided probability value of less than 0.05 was considered statistically significant and unlikely a result of chance. Patients alive at the end of the study period were censored for purposes of data analysis.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
0.001). Table 3
compares the outcomes for older and younger patients. There was no significant difference in hospital length of stay (4.3 days in those 45 years and 4.8 days in those >45 years; p = 0.359) or major morbidity (in 26 or 10% in those 45 years and in 61 or 12% in those >45 years; p = 0.548). However, there was a lower operative mortality in the younger patients (1 operative mortality, 0.4%) compared with 20 (3.8%) operative mortalities in the older than 45 years group (p = 0.005).
|
|
|
|
45 years and 4.7 days in those >45 years; p = 0.406) or major morbidity (in 12% in those 45 years and 15% in those >45 years; p = 0.431). Additionally, the 5-year overall Kaplan–Meier survival remained significantly lower in the younger group compared with the older (44% versus 57%; p = 0.029). |
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Another recent series, by Tian and associates [11] in 2003, examined the 5-year survival of 92 young patients with NSCLC. The authors reported the 5-year survival was 46% in patients younger than 40 years of age, significantly higher than that of patients older than 40 who underwent surgery. Although their 5-year survival is similar to the one reported in our series (51% compared with 46%), their cohort included patients who were clinically staged only and received neoadjuvant therapy.
Bernet and colleagues [12] presented another paper that compared outcomes between young and old patients with NSCLC. Ninety-two patients were younger than 50 and 120 were older than 70. The authors reported similar adjusted (tumor-related) survival rates at 5 years for both groups overall and for stage I, stage IIIA, squamous cell, and adenocarcinoma. This study also suffers from the same inaccuracies of staging as described above.
Our series produced several unique findings. One was a difference in histologic characterization between the older and younger cohorts. Gadgeel and coworkers [9], Tian and associates [11], and others have reported varying proportions between squamous and adenocarcinoma between the two groups. However, in our series, those younger patients had a greater proportion of non-adenocarcinoma, non-squamous NSCLC (mucoepidermoid and carcinoid). Some might argue that these patients should be excluded from the study analysis. We believe it would be a mistake to exclude these diseases, which are more frequently seen in younger patients, in a clinical study that purports to compare the differences between younger and older patients with NSCLC. However, we performed a separate analysis that excluded all of these patients from both groups. As expected, we found an even further widening in the survival advantage in the elderly patients. This was predictable as the majority of patients with mucoepidermoid and carcinoid tumors were in the younger group and patients with these types of favorable tumors had outstanding 5-year survival.
Another unique finding in our study was that younger patients presented to the thoracic surgeon later than elderly patients. Patient-bias or physician-bias may explain this finding. A younger patient is less likely to consider cancer and thus may delay being seen. Similarly, the general practitioner may place cancer on the end of the differential and therefore may delay in ordering tests such as a chest CT scan for a younger patient as compared with an older patient with a similar presentation. Other possible explanations include access to care issues, as younger patients are less likely to go for routine checkup as compared with older patients who often require more frequent visits for medication refills and laboratory work. Given these possible explanations, it is not surprising that we also observed that younger patients were more likely to be symptomatic at the time of presentation. Finally, in this series we observed that a greater proportion of younger patients were nonsmokers. We observed similar age-related findings in our previous series that evaluated NSCLC in smokers compared with never smokers [13]. This finding may add support to the claim that younger patients have a more genetically based cause of NSCLC as compared with older patients.
The strengths of this study include its methodology, especially the elimination of confounding by patients who received neoadjuvant therapy (most of whom were only clinically staged). Matching was performed 1:2 younger to older to increase power and minimize the effect of confounders. We matched for confounders such as stage. We also matched for sex because we and others have shown that women with NSCLC have a better 5-year survival than men of equal stage [14]. Additionally, the prospective database from which our cohort was extracted has a 99% follow-up rate and is comprehensive and more complete than larger national registries such as SEER. For example, all data are entered prospectively by one person and contain topic-specific variables such as smoking history and time from symptoms presentation until date seen in clinic. The biggest strength of our study, however, is the meticulous staging. Staging was performed by one physician using a single algorithm. All patients had a CT and PET scan, thus reducing the incidence of missed M1 disease. All patients had an open thoracotomy, careful palpation of the other lobes of the lung thought to be uninvolved with cancer [15], and complete thoracic lymphadenectomy (not just sampling). This ensures confirmed pathologic staging, not just guessed at or estimated clinical staging.
There are many important limitations to our study as well. The entry criteria we selected injected an inherent bias. We report only on patients who underwent resection. Hence those with stage IV disease are quite different from the vast majority of patients with stage IV NSCLC. These patients were those in whom we found pulmonary nodules of the same histologic designation in two different lobes at the time of resection or patients with known already resected brain metastases. Furthermore, our study is uni-institutional and limited geographically; thus, its findings may not be generalized to other populations.
In conclusion, we have shown that youth portends a significantly worse prognosis in patients with NSCLC. Younger patients are more likely to be symptomatic and have a greater delay in seeking thoracic surgical care. Thus, it is paramount for all health-care providers to consider lung cancer in young patients who are nonsmokers. Given the propensity for NSCLC to present in young nonsmokers, further investigation into genetic or tumor molecular characteristics are needed. These factors along with behavior factors may explain the worse survival of these younger patients with NSCLC. Finally, these data need to be considered in the treatment algorithms for younger patients. Perhaps a more aggressive multimodality approach is indicated even if younger patients undergo resection of an early staged NSCLC. In addition, a more intense investigation for metastatic disease before resection or a more careful surveillance after resection is indicated. Further genetic testing of tumors and of a patient's genome are needed.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Obviously this patient population is one that is receiving a lot of attention, and you pointed out those facts, that these patients are exposed to much less environmental toxins. So whatever reason is causing them to develop this cancer, it is probably more genetic, and so this is a great target population to begin to explore genetic etiology of cancer. But in order to be able to apply the findings genetically from this group to our older patients, we really do have to find out where they differ in their clinical presentation and in the aggressiveness of their tumor. So this sort of work is extremely important.
You had actually good fortune to report on a very large population as well; 254 young lung cancer patients is an enormous number, and really when you look at the review that is done in the paper, it is one of the largest subgroups of young patients that is published out there.
My concerns with the paper really rest on one issue of your methodology. Your initial purpose was to examine two separate, really, groups of outcomes: one was focused on patient characteristics at presentation, the other one was focused on the actual cancer outcomes in the survival. When we look at the presentation of younger cancer patients, you accurately portray how these patients present. You really need to examine an unmatched group. You looked at a time period from 1999 to 2006 and took consecutively every patient who presented under the age of 45. You then matched them on only four variables to a subgroup of all the older patients that you saw. I know how busy Cerfolio is. You probably could have found a well-matched group on stage and type of procedure who were all adenocarcinomas, if you wanted to, or you could have found all squamous cell carcinomas if you wanted to. By selecting a matched group to compare presenting symptoms, you are arbitrarily limiting the type of patients you are seeing who are older. And so I have a question about just why you chose that methodology?
The second question is in a similar way. If you are going to assess outcomes, particularly survival, and you want to make the only factor responsible for differences that you see in survival to be age, then you truly need to match on every single variable that we know impacts on survival, things like the histologic grade, things like the number of lymph nodes that were involved in N1 and N2 patients, things like the size of the tumor rather than simply calling it a stage T2, perhaps a T3, or some of these newer staging systems that have come that actually look at the centimeter size of the tumor. And so if you are going to make conclusions based on a more aggressive tumor, you really need to compare the groups much more accurately. So my second question is, how do you come to grips with that, that you didn't exactly match them on every variable we know?
DR BRYANT: Thank you, Dr Vaporciyan, for your excellent comments. My response to your first question is, yes we could have compared the younger patients to the entire older cohort; however, we chose to use a nested case-control study for several reasons. First of all, case-control and nested case-control studies often allow more power to detect subtle associations. Also, we wanted to minimize the variables of interest and wanted to minimize any potential interaction between confounders and the variables of interest. For example, we have shown in our previous work that women tend to do worse stage for stage in terms of survival compared to men. So we wanted to eliminate gender influence on our outcomes by matching.
Also, we did not match on histology specifically because we were looking for differences between the younger and the older patients, and we did not want to mask any potential differences in histology by matching on that factor. It is well known that younger patients are more likely to have carcinoid so is it really fair to eliminate those with carcinoid when performing an analysis on lung cancer in young versus old? I do not think so. Yet even when we included the carcinoids the young patients did worse.
We did not match on time to presentation only because it was a surprising finding to us. In previous literature that we reviewed, there is only one study that looked at time to presentation, and they did not find any association.
You make an excellent point in that we should consider the specific pathologic characteristics of the tumor and include those variables in our analysis. We did not evaluate the grade or the number of lymph nodes involved, and we may incorporate those items into our analysis.
DR JOHN A. HOWINGTON (Cincinnati, OH): You have excellent follow-up, and made an excellent presentation. You had a 4% mortality rate in patients under the age of 45. I know Rob is an excellent surgeon. I am assuming this is based on good follow-up and actually knowing what happened to these patients after their discharge from the hospital, but can you give us an idea of what was the cause of mortality in these under 45-year-olds?
DR BRYANT: The initial in-hospital operative mortality figure we had was very low, only 1 patient. However, when we did more aggressive follow up, this number rose considerably. Most of the deaths occurred following discharge.
DR CERFOLIO: This number seemed very high to me as well, and we are double checking it, but for now, it is the numbers that we have, so we reported it honestly. It turns out the mortality was 0.4% or 1 patient, not 4%.
Parenthetically, though, what is interesting, in my experience, is the mortality of some patients who go home who did great in the hospital. But when you have a nurse like we recently have added that can call them to ensure all is okay, my morbidity and even some mortalities after hospital discharge have increased. You call to ensure all is okay, and you find out they are dead on postoperative day 49 for whatever reason. The wife or husband doesn't know, but they wake up that morning but their spouse did not, he or she is lying there dead. That counts as an operative mortality if it occurs within 30 days, or if you are reporting 90 days, it is a 90-day mortality.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. S. Bryant and R. J. Cerfolio The analysis of a prospective surgical database improves postoperative fast-tracking algorithms after pulmonary resection. J. Thorac. Cardiovasc. Surg., May 1, 2009; 137(5): 1173 - 1179. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
