Complete Response Following Preoperative Chemotherapy for Resectable Non-Small Cell Lung Cancer*

Complete Response Following
Preoperative Chemotherapy for
Resectable Non-Small Cell Lung
Accuracy of Clinical Assessment Using the French
Trial Database
Bernard Milleron, MD, FCCP; Virginie Westeel, MD, PhD;
Elisabeth Quoix, MD; Denis Moro-Sibilot, MD; Denis Braun, MD;
Bernard Lebeau, MD; and Alain Depierre; for the French Thoracic
Cooperative Group
Background: Pathologic complete response (CR) to preoperative chemotherapy has been shown to be
a strong prognostic factor in resected non-small cell lung cancer (NSCLC). This preoperative setting
offers the opportunity to evaluate the clinical prediction of CR by investigators and an evaluation
committee (EC) using the “gold standard” pathologic examination as the reference. The only
published large randomized trial of preoperative chemotherapy (to our knowledge), the French
neoadjuvant study, constitutes an interesting database to evaluate CT scan-based CR assessment.
Study objectives and design: The French trial compared mitomycin-ifosfamide-cisplatin followed by
surgery with surgery alone in stage I (except T1N0) to IIIa resectable NSCLC. Response was
prospectively assessed in all patients receiving preoperative chemotherapy by the investigator in
charge of the patient and by an EC, and was compared with pathologic postoperative data.
Results: In the preoperative chemotherapy study, 167 patients were operated on. Nineteen patients
were found to have a pathologic CR. Only seven patients were classified as having a CR by
investigators and five patients by the EC. Evaluation of CR was correct in six of these seven cases and
in three of these five cases, respectively. Sensitivity of the CR diagnosis was 31.6% for investigators
and 15.8% for the EC. Specificities of the CR diagnosis were 99.4% and 98.8%, respectively. Positive
predictive values were 85.7% and 60%, respectively. Negative predictive values were 91.9% and
90.1%, respectively. Accuracies were 91.6% and 89.2%, respectively.
Conclusion: Investigator assessment of CR was highly predictive of pathologic CR. However, this
study showed that clinical CT scan-based assessment, whether performed by investigators or the EC,
underestimated the frequency of CR after preoperative chemotherapy in resectable NSCLC.
(CHEST 2005; 128:1442–1447)
Key words: complete response; lung cancer; non-small cell lung cancer; preoperative chemotherapy; response to
Abbreviations: CR ⫽ complete response; EC ⫽ evaluation committee; FDG ⫽ F-18 fluorodeoxyglucose; NSCLC ⫽ nonsmall cell lung cancer; PD ⫽ progressive disease; PET ⫽ positron emission tomography; PR ⫽ partial response;
SUV ⫽ standardized uptake value
verall survival and response are usual end points
O used
for clinical trials in oncology. In contrast to
the straightforward survival end point, assessment of
response is highly dependent on the accuracy of
successive tumor measurements. Response evaluation is in most cases based on imaging procedures,
*From Tenon University Hospital (Dr. Milleron), CancerEst,
Paris; J Minjoz University Hospital (Dr. Westeel), Besanc¸on;
University Hospital (Dr. Quoix), Strasbourg; Michalon University
Hospital (Dr. Moro-Sibilot), Grenoble; General Hospital (Dr.
Braun), Briey; Saint Antoine University Hospital (Dr. Lebeau),
CancerEst, Paris; and J Minjoz University Hospital (Mr. Depierre), Besanc¸on, France.
Manuscript received November 23, 2004; revision accepted
January 6, 2005.
Reproduction of this article is prohibited without written permission
from the American College of Chest Physicians (www.chestjournal.
Correspondence to: Bernard Milleron, MD, Service de Pneumologie, Hoˆpital Tenon, 4, Rue de la Chine, 75020 Paris, France;
e-mail: [email protected]
Downloaded From: on 10/21/2014
Clinical Investigations
mainly CT scan. In lung cancer, thoracic CT scan has
been shown to be superior to chest radiography.1,2
However, technical quality and especially reproducibility of CT slices at the level of the target lesions
impact on measurements. Furthermore, intraobserver and interobserver variability introduces the
most important level of uncertainty.3,4 Comparisons
between the investigator assessment and evaluation
by an external evaluation committee (EC), the latter
being considered as the reference, found differences
in the rates of tumor response.5 Reasons for disagreements included errors in tumor measurements
and in the choice of targets, and technical radiologic
problems.6 With the increasing number of preoperative chemotherapy studies in non-small cell lung
cancer (NSCLC), clinical prediction of response to
chemotherapy can now be compared to the examination of resected specimens. For the evaluation of
complete response (CR), pathologic reference is
considered as the “gold standard.”
Pathologic CR after preoperative chemotherapy
for stage IIIA NSCLC has been shown to be associated with better outcome.7 Therefore, identification
of CR is an important issue. Response evaluation
remains in most patients based on thoracic CT scan.
A small series8 of preoperative chemotherapy suggested that chest CT scan underestimated the effect
of induction treatment. To our knowledge, the
French neoadjuvant study9 is the only large published randomized trial comparing chemotherapy
followed by surgery to surgery alone in stage I to IIIa
NSCLC. This study provides a large database that
can be used to better evaluate CT scan-based CR
evaluation, and not only in stage IIIa disease but also
in stage I and II NSCLC. Therefore, we evaluated
the ability of both investigators and an EC to predict
pathologic CR in patients treated with preoperative
chemotherapy for NSCLC within the French neoadjuvant randomized trial.
Methods and Materials
Study Population
Eligibility criteria, treatment allocation, and schedule have
been extensively described in the publication of the French trial.9
Briefly, eligibility criteria were as follows: (1) pathologically
proven NSCLC, (2) resectable tumor (clinical stage I except
T1N0, stage II and IIIA), (3) operable patients, (4) age ⱕ 75
years, and (5) World Health Organization performance status
ⱕ 2. Pretreatment investigations included chest radiography,
thoracic CT scan, brain CT scan or MRI, adrenal CT scan and/or
abdominal ultrasound, fiberoptic bronchoscopy, pulmonary function tests, cardiovascular evaluation, total blood cell count and
differential, and biochemistry. Patients were randomly assigned
to surgery alone or to receive preoperative chemotherapy followed by surgery. The treatment plan is illustrated in Figure 1.
Preoperative chemotherapy consisted of two cycles at a 3-week
interval of mitomycin (6 mg/m2 at day 1), ifosfamide (1.5 g/m2 at
day 1 to day 3), and cisplatin (30 mg/m2 at day 1 to day 3). Surgery
was planned during the seventh week. In responders, two
additional chemotherapy cycles were to be delivered after surgery. Only patients included in the preoperative chemotherapy
arm who underwent surgery were considered for the present
Response Evaluation
Response to chemotherapy was prospectively evaluated in all
patients of the preoperative chemotherapy arm by the investigator in charge of the patient 1 week before surgery, with thoracic
Figure 1. The French prospective neoadjuvant trial: mitomycin (6 mg/m2 day 1), ifosfamide (1.5 g/m2
day 1 to day 3), and cisplatin (30 mg/m2 day 1 to day 3). NC ⫽ no change; P ⫽ progression.
Downloaded From: on 10/21/2014
CHEST / 128 / 3 / SEPTEMBER, 2005
CT scan and fiberoptic bronchoscopy plus bronchial biopsies. All
data were subsequently reviewed by an EC constituted of all
other investigators. A total of 38 centers participated in this
multicenter study. This evaluation by the EC was blinded with
regard to both the investigator assessment and the postoperative
pathologic findings.
CR assessment used the World Health Organization criteria.10
CR was defined as the complete disappearance of all target
lesions without any residual lesion. The absence of cancer cells in
bronchial biopsy specimens was required for clinical CR. Partial
response (PR) was defined as a ⬎ 50% decrease in tumor mass,
without progression in any target lesion or appearance of a new
lesion. Stable disease was defined as either a ⬍ 50% decrease or
a ⬍ 25% increase of tumor mass without appearance of a new
lesion. Progressive disease (PD) was defined as a ⬎ 25% increase
in tumor mass or the appearance of a new lesion.
Pathologic response was determined after examination of all
resected histopathologic materials, and classified as CR or others.
Patients whose specimens contained no cancer cells were considered as having a pathologic CR. All other possibilities were
called “others.”
Statistical Analysis
The ability of both investigators and the EC to predict
pathologic CR was measured using the calculation of sensitivity,
specificity, accuracy, and positive and negative predictive values.
Pathologic assessment was considered as the reference. The
degree of agreement between investigators and the EC was
provided for all categories of response (CR, PR, stable disease,
progression, and nonevaluable) using a ␬ coefficient.
Study Population
From June 1991 to July 1997, 373 NSCLC patients were included in the French neoadjuvant
study (187 patients enrolled in the preoperative
chemotherapy arm and 186 in the primary surgery
arm). Among the 187 patients of the preoperative
chemotherapy arm, 179 were eligible. Clinical staging resulted in 62 stage IB, 25 stage II, and 92 stage
IIIA NSCLCs. Pathologic type was squamous cell
carcinoma in 129 patients, adenocarcinoma in 30
patients, and large cell carcinoma in 20 patients.
Among these 179 patients, 167 were operated on and
constituted the database of this analysis.
Response Evaluation
Nineteen patients were found to have a pathologic
CR after thoracotomy. Of these 19 patients, 7 patients had stage I, 7 patients had stage II, and 5
patients had stage IIIA NSCLC. Pathologic type was
squamous cell carcinoma in 13 patients, adenocarcinoma in 3 patients, and large cell carcinoma in 3
Investigator and EC response assessments and
pathologic responses are displayed in Table 1. Only
seven patients were identified as having a CR by
Downloaded From: on 10/21/2014
Table 1—Investigator, EC, and Pathologic CR in the
167 Patients Operated on After Chemotherapy*
No change
7 (4.2)
5 (3)
19 (11.4)
*Data are presented as No. (%) or No.
their investigator and five patients by the EC. Thus,
13 of the 107 operated-on patients (14%) classified
by investigators as PRs and 15 patients (16.1%) of
the 93 PRs defined by the EC had in fact a CR. The
results of investigators and the EC compared with
those of pathologic assessment are detailed in Table
2. Evaluation of CR was correct in six of the seven
cases defined by investigators and in three of the five
cases of the EC. All the patients with a pathologic
CR not identified by investigators had been considered as having a PR. Among the 16 cases of pathologic CR missed by the EC, 15 cases had been
classified as PRs but 1 case had been evaluated as
stable disease. Four (three in stage IIIaN2) of the six
CRs identified by investigators and one of the three
CRs predicted by the EC occurred in stage IIIa
disease. Sensitivity of the CR diagnosis was 31.6%
for investigators and 15.8% for the EC. Specificities
of the CR diagnosis were 99.4% and 98.8%, respectively. Positive predictive values were 85.7% and
60%, respectively. Negative predictive values were
91.9% and 90.1%, respectively. Accuracies were
91.6% and 89.2%, respectively.
Comparisons between investigator and EC response evaluations are presented in Table 3. Only 2
of the 19 pathologic CRs were correctly predicted by
Table 2—Investigator Response vs Pathologic CR and
EC Response vs Pathologic CR in the 167 Patients
Operated on After Chemotherapy
Pathologic Response
No change
No change
Clinical Investigations
Table 3—Comparison of Response Evaluation Between
Investigators and EC in the 167 Patients Operated on
After Chemotherapy
Evaluation Committee
No Change
No change
both investigators and the EC. Few patients were
classified as responders by investigators and PD by
the EC or vice versa. The ␬ coefficient was
0.59 ⫾ 0.113 (mean ⫾ SD). Disagreements were
more frequent in stage I (␬ ⫽ 0.46 ⫾ 0.37) and stage
IIIA (␬ ⫽ 0.46 ⫾ 0.17) than in stage II NSCLC
(␬ ⫽ 0.85 ⫾ 0.11).
This study showed on a large database that pathologic CR was largely underestimated by both investigators and the EC, with only 6 and 3 of the 19
pathologic CRs predicted, respectively. All CRs
missed by investigators and all but one of the CRs
not identified by the EC were considered as clinical
PRs. However, investigator assessment of CR was
highly predictive of a pathologic CR. As most published studies of preoperative chemotherapy have
focused on stage III disease, it is interesting to notice
in this series, including half stages I and II, that
clinical prediction of pathologic CR was easier in
early stages.
In a phase II study11 by the Swiss Group for
Clinical Cancer Research, a similar analysis was
performed in a smaller group of patients. Among the
75 patients who underwent tumor resection after
three cycles of docetaxel-cisplatin for operable stage
IIIA pathologic N2 NSCLC, 14 pathologic CRs
(19%) were observed. Definition of pathologic CR
was different from that used in the present analysis,
also including patients with a few persistent viable
tumor cells. All tumors containing ⱖ 95% necrosis
and fibrosis were considered as pathologic CRs. Of
the 14 pathologic CRs, clinical staging assessed by
medical panels resulted in 6 CRs, 5 PRs, and 3
stabilizations. In an Italian retrospective analysis12
conducted in 76 patients operated-on for a stage III
NSCLC after chemoradiation, none of the eight
pathologic CRs were clinically predicted. Hope for
better clinical CR evaluation could be awaited from
an EC. Indeed, review of all presumed responses by
Downloaded From: on 10/21/2014
an EC, which assesses compliance with evaluation
guidelines and verifies consistency of tumor measurements, has been reported to produce more
reliable overall response evaluation.6 The present
study, with pathologic reference available, allows
evaluation of an EC for the specific situation of
pathologic CR. Our data did not confirm the superiority of EC over investigators in the assessment of
pathologic CR, with an even worse sensitivity than
that of investigators (15.8% for the EC and 31.5% for
investigators). Clinical assessment of a CR defined as
disappearance of all targets without any residual
lesions is very difficult. Most CT scan presentations
consist of residual masses impossible to distinguish
between fibrosis and necrosis only and a mixed
pattern of fibrosis, necrosis, and viable cancer cells,
as they present no specific features. For CR, a
too-rigorous review of response assessment as provided by an EC did not seem to be helpful.
Pathologic CR after preoperative chemotherapy is
significantly correlated with improved survival.7 It
requires clearance of both the primary tumor and
lymph nodes, and especially mediastinal lymph
nodes usually referred to as mediastinal downstaging. Emphasis is often placed on mediastinal lymph
node CR, which has also been shown to be a strong
prognostic factor.11,13,14 Among the 75 patients who
underwent tumor resection in the Swiss study,11 23
patients (31%) downstaged to N0 and 22 patients
(21%) downstaged to N1. Downstaging to N0 –1 was
associated with highly significantly improved survival
(61% vs 11% at 3 years, p ⫽ 0.0001). In multivariate
analysis, CR and mediastinal lymph node downstaging were the only significant prognostic factors, with
mediastinal downstaging being the most powerful
(hazard ratio, 0.22; 95% confidence interval, 0.10 to
0.49; p ⫽ 0.0003). However, assessment of overall
CR should not be neglected, as nodal downstaging is
significantly more frequent in patients with pathologic CR in the primary tumor. In the Swiss study,11
mediastinal downstaging was observed in 93% of
pathologic CRs in the primary tumor, and in only
57% of patients with less activity in the primary
tumor (p ⫽ 0.013). Standard treatment still needs to
be defined for stage III disease, and identification of
mediastinal lymph node clearance and of complete
responders might become crucial to optimize the
indications of combined modality therapy. In the
Intergroup Trial 0139,14 whose preliminary results
were first presented at the 2003 meeting of the
American Society of Clinical Oncology, patients with
a pIIIaN2 NSCLC were randomized between induction concomitant chemoradiation followed by surgery and chemotherapy, or further radiotherapy and
consolidation chemotherapy. Three-year survival was
not improved in the trimodality arm (38% vs 33%)
CHEST / 128 / 3 / SEPTEMBER, 2005
but was approximately 50% for operated-on patients
who had achieved a mediastinal pathologic CR.
Clinical evaluation of response being insufficient,
other procedures are being tested and particularly
positron emission tomography (PET) scanning. In
patients with aggressive non-Hodgkin lymphoma,
F-18 fluorodeoxyglucose (FDG)-PET scan appears
to be useful for the diagnosis of CR.15 In a prospective study of 70 patients, none of the 33 patients who
showed persistent abnormal FDG uptake achieved a
durable complete remission, whereas 31 of 37 patients with normal scan findings remained in complete remission with a median follow-up of 1,107
days. In solid tumors, it is uncertain whether FDG
uptake may predict a CR. In breast cancer, 10
patients underwent FDG-PET scanning before definitive breast surgery.16 No abnormal uptake at the
primary tumor site was visualized in any patient,
whereas 9 of the 10 patients had residual invasive
carcinoma at operation, ranging from 2 to 20 mm in
maximum dimension. Similarly, sensitivity of PET
for evaluating a CR in patients with ovarian or
peritoneal carcinoma is very low.17 Several studies18,19 evaluating PET scan restaging after preoperative chemotherapy or chemoradiation in NSCLC
showed a high sensitivity (97 to 100%) but limited
specificity (58 to 67%) for detecting residual primary
tumors, and good specificity (75 to 99%) but poor
sensitivity (50 to 61%) for lymph nodes. One of these
studies19 compared PET and CT scan restaging in 34
patients treated with chemotherapy or chemoradiotherapy for NSCLC. PET scanning was shown to be
more specific (67% vs 0) for detecting residual tumor
and more sensitive (50% vs 30%) for N2 disease than
CT scan.19 The same authors20 stated that there was
a tight correlation between maximum standardized
uptake value (SUV) and pathologic CR with a sensitivity of 90%, a specificity of 100%, and an accuracy
of 96% in predicting histopathologic CR when maximum SUV is decreased by ⱖ 80%. However, Port et
al21 with a cut-off value of 50% decrease in SUV,
showed that positive and negative predictive values
were rather low as well in tumor and in nodes.
Reproducibility of SUV from one center to another
may also account for some of the discrepancies in the
interpretation of the results. Although better response evaluation can be obtained with PET scanning, remediastinoscopy remains the most accurate
procedure for the evaluation of mediastinal response, as it provides histologic evidence. However,
clinical response evaluation cannot be neglected, as
remediastinoscopy remains an invasive procedure
and gives no information on the primary tumor,
which can be of great importance in some cases of
marginally resectable tumors.
Downloaded From: on 10/21/2014
This study showed that investigator assessment of
CR was highly predictive of pathologic CR. However, this study showed that clinical CT scan-based
assessment, whether performed by investigators or
the EC, underestimated the frequency of CR after
preoperative chemotherapy in resectable NSCLC. If
selection of patients for any treatment should depend on CR evaluation, clinical assessment either by
the investigator in charge of the patient or by an EC
could not be relied on.
1 Dajczman E, Hanley J, Lisbona A, et al. Comparison of
response evaluation in small cell lung cancer using computerized tomography and chest radiography. Lung Cancer 1994;
11:51– 60
2 Pujol JL, Demoly P, Daures JP, et al. Chest tumor response
measurement during lung cancer chemotherapy: comparison
between computed tomography and standard roentgenography. Am Rev Respir Dis 1992; 145:1149 –1154
3 Herschorn S, Hanley J, Wolkove N, et al. Measurability of
non-small-cell lung cancer on chest radiographs. J Clin Oncol
1986; 4:1184 –1190
4 Quoix E, Wolkove N, Hanley J, et al. Problems in radiographic estimation of response to chemotherapy and radiotherapy in small cell lung cancer. Cancer 1988; 62:489 – 493
5 Gwyther SJ, Aapro MS, Hatty SR, et al. Results of an
independent oncology review board of pivotal clinical trials of
gemcitabine in non-small cell lung cancer. Anticancer Drugs
1999; 10:693– 698
6 Thiesse P, Ollivier L, Di Stefano-Louineau D, et al. Response
rate accuracy in oncology trials: reasons for interobserver
variability. Groupe Francais d’Immunotherapie of the Federation Nationale des Centres de Lutte Contre le Cancer.
J Clin Oncol 1997; 15:3507–3514
7 Pisters KM, Kris MG, Gralla RJ, et al. Pathologic complete
response in advanced non-small-cell lung cancer following
preoperative chemotherapy: implications for the design of
future non-small-cell lung cancer combined modality trials.
J Clin Oncol 1993; 11:1757–1762
8 Schaefer-Prokop C, Prokop M. New imaging techniques in
the treatment guidelines for lung cancer. Eur Respir J Suppl
2002; 35:71s– 83s
9 Depierre A, Milleron B, Moro-Sibilot D, et al. Preoperative
chemotherapy followed by surgery compared with primary
surgery in resectable stage I (except T1N0), II, and IIIa
non-small-cell lung cancer. J Clin Oncol 2002; 20:247–253
10 Miller AA, Hargis JB, Lilenbaum RC, et al. Phase I study of
topotecan and cisplatin in patients with advanced solid tumors: a cancer and leukemia group B study. J Clin Oncol
1994; 12:2743–2750
11 Betticher DC, Hsu Schmitz SF, Totsch M, et al. Mediastinal
lymph node clearance after docetaxel-cisplatin neoadjuvant
chemotherapy is prognostic of survival in patients with stage
IIIA pN2 non-small-cell lung cancer: a multicenter phase II
trial. J Clin Oncol 2003; 21:1752–1759
12 Margaritora S, Cesario A, Galetta D, et al. Ten year experience with induction therapy in locally advanced non-small cell
lung cancer (NSCLC): is clinical re-staging predictive of
pathological staging? Eur J Cardiothorac Surg 2001; 19:894 –
Clinical Investigations
13 Bueno R, Richards WG, Swanson SJ, et al. Nodal stage after
induction therapy for stage IIIA lung cancer determines
patient survival. Ann Thorac Surg 2000; 70:1826 –1831
14 Albain KCS, Rush VR, Turrisi AT, et al. Phase III comparison
of concurrent chemotherapy plus radiotherapy (CT/RT) and
CTRT followed by surgical resection for stage IIIA(pN2)
non-small cell lung cancer (NSCLC): initial results from
Intergroup Trial 0139 (RTOG 93– 09) [abstract]. Proc Am
Soc Clin Oncol 2003; 22:2497
15 Spaepen K, Stroobants S, Dupont P, et al. Early restaging
positron emission tomography with (18)F-fluorodeoxyglucose
predicts outcome in patients with aggressive non-Hodgkin’s
lymphoma. Ann Oncol 2002; 13:1356 –1363
16 Burcombe RJ, Makris A, Pittam M, et al. Evaluation of good
clinical response to neoadjuvant chemotherapy in primary
breast cancer using [18F]-fluorodeoxyglucose positron emission tomography. Eur J Cancer 2002; 38:375–379
17 Rose PG, Faulhaber P, Miraldi F, et al. Positive emission
tomography for evaluating a complete clinical response in
Downloaded From: on 10/21/2014
patients with ovarian or peritoneal carcinoma: correlation
with second-look laparotomy. Gynecol Oncol 2001; 82:17–21
Ryu JS, Choi NC, Fischman AJ, et al. FDG-PET in staging
and restaging non-small cell lung cancer after neoadjuvant
chemoradiotherapy: correlation with histopathology. Lung
Cancer 2002; 35:179 –187
Cerfolio RJ, Ojha B, Mukherjee S, et al. Positron emission
tomography scanning with 2-fluoro-2-deoxy-d-glucose as a
predictor of response of neoadjuvant treatment for nonsmall cell carcinoma. J Thorac Cardiovasc Surg 2003; 125:
938 –944
Cerfolio R, Bryant A, Winokur T, et al. Repeat FDG-PET
after neoadjuvant therapy is a predictor of pathologic response in patients with non-small cell lung cancer. Ann
Thorac Surg 2004; 78:1903–1909
Port JL, Kent MS, Korst RJ, et al. Positron emission tomography scanning poorly predicts response to preoperative
chemotherapy in non-small cell lung cancer. Ann Thorac Surg
2004; 77:254 –259, discussion 259
CHEST / 128 / 3 / SEPTEMBER, 2005