| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;83:409-418
© 2007 The Society of Thoracic Surgeons
a Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, Minnesota
b Division of General Thoracic Surgery, Mayo Clinic College of Medicine, Rochester, Minnesota
c Section of Biostatistics, Mayo Clinic College of Medicine, Rochester, Minnesota
Accepted for publication August 28, 2006.
* Address correspondence to Dr Nichols, Division of General Thoracic Surgery, Mayo Clinic, 200 First St, SW, Rochester, MN 55905 (Email: nichols.francis{at}mayo.edu).
Presented at the Forty-first Annual Meeting of The Society of Thoracic Surgeons, Tampa, FL, Jan 24–26, 2005.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
METHODS: We prospectively followed 1,361 consecutive patients with nonsmall-cell lung cancer who underwent complete surgical resection at our institution from January 1997 to December 2001. Only patients having recurrent cancer were included in the analysis. Multivariable Cox proportional hazards models were used to evaluate the effect of prognostic factors on postrecurrence survival.
RESULTS: Follow-up was achieved in 1,073 patients, and recurrent cancer developed in 445. Complete information was available on 390 patients for analysis. There were 262 men and 128 women. Median age at time of recurrence was 69 years. Median time from surgical resection to recurrence was 11.5 months, and median postrecurrence survival was 8.1 months. Recurrence was intrathoracic in 171 patients, extrathoracic in 172, and a combination of both in 47. Treatments after recurrence included surgery in 43 patients, chemotherapy in 59, radiation in 73, and a combination in 96. All patients who received treatment survived longer than those who received no treatment. Preoperative chemotherapy and postoperative radiotherapy for the primary lung cancer, poor Eastern Cooperative Oncology Group Performance Status, decreased disease-free interval from initial resection to recurrence, symptoms at recurrence, and certain location of recurrence significantly decreased postrecurrence survival.
CONCLUSIONS: In our experience, treatment for recurrent nonsmall-cell lung cancer significantly prolongs survival. Various treatment modalities including surgery should be considered in patients with postoperative recurrent nonsmall-cell lung cancer.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
The postrecurrence survival (PRS) of surgically treated patients has not been the focus of much attention. Relatively few studies have evaluated the prognostic factors associated with survival after the recurrence of lung cancer. Chemotherapy and radiation therapy are commonly accepted treatment options for recurrent lung cancer, whereas surgical resection is limited to a relatively few patients. The purpose of this study was to identify prognostic factors associated with PRS in surgically treated NSCLC patients who ultimately have recurrent lung cancer.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Information abstracted from the medical record for each patient included age, sex, race, smoking status, history of tobacco exposure, other medical conditions, symptoms at presentation, and Eastern Cooperative Oncology Group Performance Status (ECOG-PS). For the initial diagnosis of primary NSCLC, date of diagnosis, clinical stage, histolopathologic type and grade of tumor, type of surgical resection, and the use of either neoadjuvant or adjuvant therapy were recorded. Histology classification was according to the World Health Organization Classification of Tumors, third edition [8]. Pathologic stage was recorded for patients not receiving neoadjuvant therapy. For patients receiving neoadjuvant therapy, both the clinical and pathologic stages were recorded, and the highest of the clinical or pathologic T and N factors were combined in the analysis. The TNM stage was classified according to the International System for Staging Lung Cancer [9].
Follow-up visits, procedures, and treatments for recurrent NSCLC were provided primarily at Mayo Clinic but did include other medical institutions. Where necessary, there was a systematic request and review of outside medical records when cancer recurrence or progression was reported as well as communication with the patient’s other health care providers.
Recurrent NSCLC was diagnosed through physical examination and diagnostic imaging of lesions consistent with recurrent lung cancer. Histologic confirmation of the diagnosis was made when clinically feasible. The date of recurrence was defined as the date of histologic proof or, in cases diagnosed based upon clinical evidence, the date of recognition of the recurrent disease by the attending physician. The evidence of recurrent disease was abstracted from Mayo Clinic medical records or other reliable sources. Additional sources of data collection included lung cancer follow-up questionnaires, tumor registry questionnaires routinely sent to the patient from Mayo Clinic, and correspondence letters or copied medical records received from outside clinicians. When the presence or absence of recurrent disease was documented at another institution, the data were included only when the information was specific and considered reliable. The criteria proposed by Martini and Melamed [10] were utilized for differentiating recurrent NSCLC from a new primary lung cancer. Patients were included in the analysis of PRS only when specific data were available for the exact anatomic site of recurrence and any treatment given throughout the follow-up period. Recurrence was considered local if it initially occurred in an anatomically contiguous area or in the regional lymph nodes of the initial primary lung cancer. Ipsilateral recurrent lung cancer not meeting this criterion was considered distant recurrence.
A single recurrent focus involving adjacent anatomic sites was not considered a multisite recurrence unless each site was treated separately, and the initial site of recurrence was classified as the single site most extensively involved with recurrent tumor. Finally, patients in whom multiple foci of initial recurrences involved multiple sites within the chest (eg, lung parenchyma and mediastinum or chest wall) were categorized as all other chest recurrences and not as multiple organ recurrences.
Treatment for recurrent disease was recorded for any surgical procedure, chemotherapy, and radiation therapy, and for combinations of these modalities. For analytical purposes, stereotactic radiosurgery for brain metastases was considered a surgical treatment. Data were collected for the initial and all subsequently diagnosed lung cancer recurrences.
To standardize the quality of follow-up information across the entire cohort of 1,361 patients with completely resected NSCLC, only patients with follow-up information complete as of December 31, 2003, were included in the analysis.
Statistical Analysis
The Kaplan-Meier method was used for calculation of the survival rates, and differences in survival were determined by log-rank analysis [11, 12]. The relative risk (RR) of death and its 95% confidence interval (CI) was calculated using the Cox proportional hazards model [13]. Demographic and clinical characteristics were initially tested for their association with PRS in the univariate analysis, and then all reported characteristics were considered in a multivariable Cox proportional hazards regression model to determine which demonstrated the strongest associations with PRS. A forward stepwise selection procedure was implemented, with a p value threshold of 0.05 for inclusion in the final model.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Recurrent NSCLC developed in 445 patients (41%). In 390 patients (88%), detailed information was available for the date and site of recurrence and the treatment provided, and these patients were included in the analysis of PRS. There were 262 men and 128 women. Median age at time of initial recurrence (IR) was 69 years (range, 33 to 90). At initial surgical resection, 76 patients were stage IA, 84 were IB, 24 were IIA, 64 were IIB, 100 were IIIA, and 42 were IIIB. The median disease-free interval between initial lung cancer resection and IR was 11.5 months (range, 10 days to 80.7 months). The median length of survival after initial lung cancer resection was 24.8 months (range, 1.6 to 81.7). One-year, 3-year, and 5-year survivals after initial surgical resection were 76%, 31%, and 12%, respectively.
For the 390 patients, PRS ranged from 2 days to 53.6 months. The median PRS was 8.1 months, and the 1- and 2-year PRS rates were 37% and 17%, respectively. Initial recurrence was intrathoracic in 171 patients, extrathoracic in 172, and a combination of both in 47 (Table 1). Seventy-five patients (19%) had their IR in two or more separate (multiple) anatomic locations. Among patients with combined intrathoracic and extrathoracic IR, the most common combination was lung and bone in 23 patients, and lung and liver in 14. The 1-year PRS rates were 50% for patients with intrathoracic recurrence, 26% for extrathoracic recurrence, and 28% for combined intrathoracic and extrathoracic recurrences (p < 0.01; Fig 1).
|
|
The demographic and clinical characteristics of the 390-patient cohort are shown in Tables 2 and 3.
These variables were tested for their association with PRS. As shown in Table 2, low ECOG-PS, asymptomatic presentation, patients who were stage I, and adenocarcinoma histopathology at initial resection had an improved PRS. Decreased PRS was observed among patients with pneumonectomy, neoadjuvant chemotherapy, or adjuvant radiation therapy. Similarly, in Table 3, both a low ECOG-PS and asymptomatic presentation at IR resulted in improved PRS. Patients with a disease-free interval of less than 1 year and patients with multiple recurrent foci had decreased PRS. The anatomic site of IR was associated with PRS (Table 3). Patients with IR confined to the lung or liver alone demonstrated the longest PRS of 15.5 and 4.2 months, respectively. When IR was treated with surgery alone, the median PRS was 22.9 months and 2-year PRS was 49%.
|
|
|
Surgical treatment was the initial treatment for the recurrent lung cancer in 76 patients (19.5%). In these patients, IR was limited to the brain in 31 patients, lungs in 23, and other anatomic sites in 22. Tables 5, 6, and 7
show the results of surgical and nonsurgical treatment in 5 anatomic sites. Adjusted RR for treatment groups were calculated and compared with patients who received no treatment. For IR limited to the lungs or brain, PRS was evaluated for both solitary and multiple foci recurrences.
|
|
|
Twenty-two patients had surgical resection for initially recurrent lung cancer in sites other than the lung or brain alone (Table 7). Table 8 shows the outcomes for 11 of those patients who had surgical resection of the recurrence with a curative intent. Seven patients were alive without evidence of new recurrence at last follow-up.
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Our results demonstrating the presence of symptoms as a predictor of poor PRS is in agreement with those from Westeel and colleagues [14] and Walsh and associates [15]. Seventy-four percent of our patients without symptoms at IR had disease confined within the chest. Conversely, 72% of patients with symptoms at IR had an extrathoracic recurrence. In a survival analysis of inoperable lung cancer patients testing a previously developed prognostic index, Wigren [16] demonstrated that the existence of general or metastatic symptoms resulted in a fourfold increase in risk when compared with presence of local or no symptoms. In our patients, because of the superior median survival with an intrathoracic IR and the lack of symptoms in the majority of these patients, we believe that the presence of existing symptoms and initial site of IR are both important predictors of PRS.
Interestingly, administration of neoadjuvant chemotherapy or adjuvant radiation was the only factor that predicted PRS before cancer recurred. Authors from the National Kyushu Cancer Center (Fukuoka, Japan) have repeatedly reported reduced PRS in patients treated with adjuvant chemotherapy or radiation therapy; however, the reason for this finding remains unclear [17–19]. Similarly, our patients treated with neoadjuvant chemotherapy or adjuvant radiation showed decreased PRS, but also demonstrated a higher likelihood of developing recurrence in the brain. Meanwhile, these patients showed unfavorable PRS regardless of its initial presentation to the brain when compared with patients not receiving additional therapy. Although frequent brain recurrence may be contributing to the worse PRS after multimodality treatment, it does not seem to explain this phenomenon entirely. On the other hand, TNM stage captures the extent of disease progression and malignant potential of primary lung cancer comprehensively; therefore, it is shown to be the most powerful prognostic tool in NSCLC [20]. As administration of neoadjuvant or adjuvant therapy is highly correlated with primary lung cancer stage, it is indicative that these additional treatments surrogated stage as a significant predictor in the multivariate analysis. In fact, when these additional treatments were excluded from the list of predictors, stage was a highly significant predictor in the final model.
Walsh and colleagues [15] demonstrated that complete surgical resection or high-dose radiotherapy with curative intent significantly prolonged PRS in NSCLC. Yoshino and colleagues [19] demonstrated that surgical resection of recurrent foci significantly prolonged survival in patients with recurrence to distant organs. Finally, Yano and coworkers [21] emphasized the benefit of local control with radiation therapy in the treatment of locally recurrent NSCLC. Our study shows that treatment, whether it was surgery or combination chemotherapy with radiation, significantly improved PRS over both no treatment and radiation alone. Our results are concordant with the prior studies in that our treatment strategies included modalities aimed at local control of recurrent disease, specifically, surgical resection or radiotherapy. For IR, surgical resection alone had a favorable PRS and low RR. However, we also showed that systemic chemotherapy played an important role in the treatment of recurrent NSCLC. Chemotherapy when combined with radiation was associated with favorable PRS, and also was one of the treatment groups with a low RR (Table 4). Furthermore, when we modeled treatment as the three major categories of surgery, chemotherapy, and radiation instead of the five treatment groups in Table 4, all three demonstrated a significantly decreased RR, namely, 0.3, 0.4, and 0.6, respectively, when compared with no treatment (RR = 1.0, reference; p < 0.01).
Objective evidence supporting the role of surgical resection in recurrent lung cancer is limited. Uncertainty exists in differentiating a second primary lung cancer from recurrent lung cancer. That is particularly true if the histopatholgy of the recurrence is initially lacking. Surgical resection is frequently the preferred treatment for a solitary metachronous lung cancer [22–24]. Our study demonstrated a similar 2-year PRS of approximately 70% in surgically treated patients with solitary and multiple lung recurrences. Surgical treatment of lung recurrences suggested a survival benefit with decreased adjusted RR when compared with nonsurgical treatments (p = 0.08). Considering these results, we believe that surgical resection should be considered a reasonable treatment option for appropriate patients with IR limited to the lungs even when multiple recurrent foci are present. We also observed that surgical resection significantly prolonged PRS in patients with recurrent disease confined to the chest, other than lung only (p = 0.02).
Many studies have shown that surgical resection of solitary extracranial metastases may result in long-term survival [19, 25–28]. Favorable survival was observed in sporadic cases of solitary recurrence to extrathoracic lymph nodes (Table 8). Surgical resection should, therefore, be considered when complete resection can be achieved.
Our results showed that surgical treatment provided prolonged median PRS for both solitary and multiple brain recurrences. While statistical significance was borderline (p = 0.07), the adjusted RR showed a strong trend toward prolonged PRS in the surgically treated group when compared with the nonsurgically treated group, and therefore, surgical resection of intracranial metastasis both solitary and multiple appears reasonable.
There are limitations to our study. As often happens in an observational cohort study, follow-up was not available for all eligible patients. Twenty-one percent of the initial eligible cohort lacked complete information as of December 31, 2003. Although baseline characteristics did not suggest a major disparity between patients with and without complete follow-up, excluding these patients may have affected the representativeness of the study group. Patients with obvious progression of recurrent disease may have received more medical attention and potentially formed a biased study group of patients in more serious medical condition. Conversely, severe illness may have disabled a subset of patients in such a way that they did not receive care from their primary healthcare providers, resulting in less follow-up information. Finally, a lack of uniformity in follow-up standards for all patients might have influenced the detection of recurrent disease and, therefore, the accurate estimation of PRS.
Randomized clinical trials will provide the most reliable evidence regarding the benefit of treatment in recurrent NSCLC; however, difficulty with patient recruitment and ethical concerns limit studies evaluating surgical interventions. Under the current circumstances, we believe that our results will add to the existing knowledge regarding this issue.
In conclusion, high ECOG-PS, symptoms at time of recurrent lung cancer diagnosis, decreased disease-free interval, and initial site of recurrence were associated with decreased PRS. Patients who fail to survive IR after receiving neoadjuvant chemotherapy or adjuvant radiation demonstrate a significantly worse PRS. Surgical resection is beneficial in the treatment of recurrent lung cancer when disease is limited to the lung alone, other chest sites, and brain. Surgery, chemotherapy, and radiation therapy individually and in combinations prolong PRS and should be considered as part of multimodality treatment when feasible.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
DR SUGIMURA: No, we did not evaluate pulmonary function as a predictor of postrecurrence survival in any way in this study.
DR CERFOLIO: But you did if you were going to re-resect them?
DR SUGIMURA: Yes, patients were carefully assessed for pulmonary function before re-resection, but the influence of pulmonary function was not evaluated against postrecurrence survival in the study.
DR SUDISH C. MURTHY (Cleveland, OH): Is it fair to really compare someone who got no treatment to someone who was treated? Don’t you think there’s a significant bias as to why someone was treated and why someone else wasn’t? The performance status of the patients who weren’t treated must have been quite poor. So you’re really comparing someone who is fit versus someone who is not fit, and might that not explain much of your results?
DR SUGIMURA: We believe that your point is critical to the success of this study. As treatment was not selected in a prospective manner, there certainly is a difference in the groups receiving different treatments. But with our prospective observational study design, we collected baseline and follow-up data prospectively for all patients with equal quality. By using this approach, we think we were able to make a valid estimate of the important predictors of postrecurrence survival, and by adjusting for those predictors, we believe that the effects of treatment were also estimated in a valid fashion for patient groups who received different treatment.
DR MURTHY: Have you used propensity matching to equalize your groups? I can understand if you’ve propensity-matched your groups and you have a group of patients who opted to not have therapy because they were going to rely on nonsurgical therapies and a similar group that opted for surgery. But unless you’ve done that, it is very difficult to conclude that you can actually make a fair apples-to-apples comparison among your groups.
DR SUGIMURA: We did not do any kind of matching. We wanted to evaluate the predictors of postrecurrence survival in our consecutive series of patients. Therefore, we did not do any matching that would limit our study group.
DR HARVEY P. RUBIN (Stamford, CT): I was wondering if you included bronchoalveolar carcinoma (BAC) among your recurrent lung cancers, and, if so, how many of those were included? I ask those questions because, in my experience, and I think that of others, it’s almost a different disease. It so often spares lymph nodes and so often has a significant incidence of recurrent lesions that can be handled very nicely, in my experience, with just a wedge resection.
DR SUGIMURA: We included all cases of pure BAC as well as cases with BAC proportionately seen within adenocarcinoma. We considered excluding these cases because they do act differently; however, we did not do so in this study. We chose to adjust for cell type as a predictor for postrecurrence survival.
DR DANIEL L. MILLER (Atlanta, GA): I enjoyed your presentation, but I think for a lot of us here, it’s an incredible amount of data. I think we need Dr Blackstone here to help us analyze the statistical analysis.
One thing I was a little concerned about, even though this was a prospective database, you only had about a 25% complete follow-up on these patients who you actually reported today, and I was a little surprised at that. My big question is that I am very surprised that the group of patients who underwent wedge excision did not fall out as a negative predictor value, and, a lot of times, we do not consider that as a true oncologic procedure. I know you count it as an R0, which at the time you would, but I’m just very amazed that that did not fall out, and I wonder if you could comment about it. I know it only represented 15%, but I think when you’re talking about a survival advantage of 12% or 8%, that would make a big difference.
DR SUGIMURA: Regarding cases that received wedge resections for the initial lung cancer, I believe that there might be a bias as well in the selection of cases that received lesser resections. Patients with poor PS may have been treated with limited resection, but on the other hand, patients who had a less invasive primary lung cancer must have received those lesser resections. I think these multiple factors influenced postrecurrence survival, resulting in a survival advantage for those who received limited or lobar resections over pneumonectomies.
For the follow-up rate, follow-up was achieved in more than 25%. It was achieved in 79% of all intended patients, and complete information on recurrence and postrecurrence survival was available on 88% of those patients with follow-up.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. Nakagawa, N. Okumura, K. Ohata, H. Igai, T. Matsuoka, and K. Kameyama Postrecurrence survival in patients with stage I non-small cell lung cancer Eur. J. Cardiothorac. Surg., September 1, 2008; 34(3): 499 - 504. [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 |
