Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print September 27, 2006, 10.1148/radiol.2412051185
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2412051185v1
241/2/554    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jennings, S. G.
Right arrow Articles by Dowdeswell, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Jennings, S. G.
Right arrow Articles by Dowdeswell, I.
(Radiology 2006;241:554-563.)
© RSNA, 2006

Thoracic Imaging

Distribution of Stage I Lung Cancer Growth Rates Determined with Serial Volumetric CT Measurements1

S. Gregory Jennings, MD, Helen T. Winer-Muram, MD, Mark Tann, MD, Jun Ying, PhD and Ian Dowdeswell, MD

1 From the Department of Radiology, Indiana University School of Medicine, 950 W Walnut St, Room E124, Indianapolis, IN 46202 (S.G.J., H.T.W., M.T., J.Y.); and Department of Medicine, Richard L. Roudebush VA Medical Center, Indianapolis, Ind (I.D.). Received July 14, 2005; revision requested September 21; revision received October 19; accepted November 26; final version accepted January 2, 2006. Address correspondence to S.G.J. (e-mail: sajennin{at}iupui.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Purpose: To retrospectively determine the distribution of stage I lung cancer growth rates with serial volumetric computed tomographic (CT) measurements.

Materials and Methods: This study was institutional review board approved and HIPAA compliant. The informed consent requirement was waived. Patients (n = 149) with stage I lung cancer who underwent two pretreatment CT examinations 25 or more days apart were identified. At the first and last examinations, tumor perimeters were manually inscribed by using software tools and the cross-sectional area was calculated. To calculate tumor volume, the summed areas were multiplied by the section increment and a formula was applied to reduce partial volume effects. Doubling time (DT) was calculated by using the volume and interscanning interval. The percentages of tumors that would surpass volume increase thresholds of 5%–25% for detectable growth at different time intervals were calculated. Age at diagnosis was compared with the reciprocal of DT, time interval between CT examinations, and initial tumor volume by using Pearson correlation. P < .05 denoted statistical significance.

Results: Lung cancer was stage IA in 99 patients and stage IB in 50. Median patient age was 72 years, and median interscanning interval was 130 days. Median tumor volumes were 3000 and 6213 mm3 at the first and last examinations, respectively. Median DT was 207 days; 21 tumors did not increase in volume between examinations. The interscanning interval required for 90% of growing tumors to surpass the growth threshold ranged from 8 weeks (5% threshold) to 37 weeks (25% threshold). Fifty-three percent of growing tumors would surpass the 25% threshold at 8 weeks, and 95% would surpass it at 1 year. Age at diagnosis was negatively correlated with growth rate (P = .047); there was no correlation between growth rate and either age at diagnosis or interscanning interval.

Conclusion: At serial volumetric CT measurements, there was wide variability in growth rates. Some biopsy-proved cancers decreased in volume between examinations.

© RSNA, 2006


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Lung cancer survival rates have been dismal (approximately 15% of cases) for many decades, but survival improves dramatically (60%) when this cancer is diagnosed at stage I (1,2). The advent of computed tomographic (CT) screening for lung cancer during the past decade has rekindled interest in the natural behavior of early lung cancers (3). The concept of screening CT is based on the theory that screening may help to increase the percentage of tumors detected before they metastasize (3–5) and decrease the size at which the tumor is first detected. Patients with stage IA tumors (diameter < 3 cm) have better survival rates than do patients with stage IB tumors (1). One result of screening CT is the detection of many small nodules that may or may not represent cancer (6). Better knowledge of the growth rates of cancers may lead to revisions in the current recommendations for CT observation of these nodules.

Many investigators have estimated the growth rates of lung cancers by using chest radiography (7–11). In (to our knowledge) the largest series (67 patients), the median time required for each lung cancer to double in volume—that is, the median doubling time (DT)—was 120 days (range, 30–490 days) (9). However, the accuracy of this estimate may be compromised, as these investigators used the mean diameter measured on each radiograph and assumed the cancer was spherical to calculate volume growth.

There have been few reports on the use of CT to measure the growth rates of lung cancers (12,13). To our knowledge, only one series of serial CT measurements—involving 50 stage I lung cancers and performed by using volumetric techniques—has been reported (14). The authors observed a DT range of 32 days (very fast growth) to –26 711 days (essentially no growth), with a median DT of 181 days.

All of these study results show that lung cancer DTs can be longer than anticipated on the basis of previously reported chest radiography literature. The reasons are unclear but may be related to the greater sensitivity of CT for detection of small, possibly slow-growing tumors (13). A current recommendation for follow-up of indeterminate nodules—if positron emission tomography (PET) and/or fine-needle aspiration biopsy is not performed—is repeat CT at 3, 6, 12, and 24 months, with the assumption that nodules that do not show growth by 24 months are benign (15). However, this schedule is based on radiographic—not CT—observations, and its scientific basis has been questioned (16). The optimal frequency of follow-up CT examinations of indeterminate nodules to detect growth is not known and is dependent on the range of growth rates of lung cancers. Because software that enables rapid and accurate semiautomated or automated measurement of tumor volumes is becoming widely available, better knowledge of the distribution of lung cancer growth rates is needed so that radiologists and clinicians can improve their ability to interpret measurement results.

At the Richard L. Roudebush VA Medical Center, many patients with lung cancer undergo serial CT before (and even after) receiving a diagnosis but before undergoing treatment. Reasons for additional CT examinations include biopsy localization, radiation therapy planning, and/or follow-up of an indeterminate nodule later diagnosed as cancer. In addition, a patient may refuse to undergo therapy, in which case the cancer will be followed up with CT. For these reasons, we have a large number of patients with stage I lung cancer with serial CT examinations available for volumetric measurement. Thus, the purpose of our study was to retrospectively determine the distribution of stage I lung cancer growth rates with serial volumetric CT measurements.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Patients
This study was institutional review board approved and Health Insurance Portability and Accountability Act compliant. The requirement for informed patient consent was waived. From our tumor registry and weekly chest conference reports, one author (S.G.J.) identified all patients who had received a diagnosis of stage I lung cancer between February 1996 and June 2004. The medical records of each patient were reviewed, and any patient who previously had received a diagnosis of primary lung cancer or cancer metastasis to the lung in the preceding 5 years was excluded. Documented patient information included age at diagnosis, date of diagnosis, cancer location and stage, type of treatment, date treatment was initiated, and method of lung cancer diagnosis. All chest CT examinations performed before the initiation of treatment and documented in our departmental archives (digital archiving began in February 1996) were identified. Only the examinations performed by using single-breath-hold spiral CT were included. Examinations that were not performed during the patient's single breath hold, in which there was noticeable motion artifact, and in which the entire tumor either was not imaged or could not be delineated were excluded.

For all remaining CT examinations, the examination date, section width, and section increment were noted, and the imaging data of all patients who had undergone at least two pretreatment examinations performed 25 or more days apart with the same scanner were examined for study inclusion. For examinations that included multiple section increments, only the images from the smallest increment that displayed the full craniocaudal extent of the tumor were used. The rationale for using 25 days as the minimum interval to detect lung tumor growth has been previously described (14,17,18).

Of 393 patients who had lung cancer diagnosed at stage I and were followed up at our institution during the study period, 136 did not undergo at least two chest CT examinations. In another 69 patients, all CT examinations were performed within an interval of less than 25 days. In another 19 patients, part of the tumor margin was obscured or the entire tumor was not imaged. In yet another 11 patients, the examinations were performed with different scanners. In nine other patients, not all of the images could be retrieved from the departmental archives. No patients were excluded because of motion artifacts. The remaining 149 patients (148 men, one woman; median age, 72 years; range, 43–87 years) met our entry criteria and constituted the study population. Data from some of these patients have been previously published (14,19,20).

CT Technique and Image Interpretation
All CT examinations were performed with a Picker PQ 2000 scanner (Marconi Medical Systems, Cleveland, Ohio) by using 120 kVp, 200 mAs, and a pitch of 1.5. Each CT image was downloaded from the archives to a video display system (MGD 521; BarcoView, Duluth, Ga) with 2500 x 2000-pixel monitors. All images were displayed at lung settings (window width, 2000 HU; window level, –700 HU). Image viewing and manipulation were controlled with Radworks 5.1 software (Applicare Medical Imaging, Zeist, the Netherlands), which allows the reader to draw lines through and perimeters around regions of interest. The software then automatically calculates the line and perimeter lengths and the area enclosed by the perimeter.

For each patient, the first and last CT examinations among the qualifying examinations were identified. The images were placed in random order, and each was then presented to one board-certified radiologist (H.T.W.) with 20 years specialized experience in chest imaging and 1 year of experience in using the image viewing and manipulation software. On each section containing tumor, the radiologist drew a line around the perimeter of the tumor.

The volume (V, in cubic millimeters) of each tumor was calculated by adding the measured cross-sectional areas (A1, A2, ...AN) and multiplying the sum by the section increment (I): V = I(A1 + A2 + ...AN), where N is the number of sections containing tumor, with the length of the z-axis equaling I · N. Additional details of the volume calculation method have been previously published (14). An equation to reduce the volumetric measurement errors due to changing section widths and tumor sizes between the CT examinations was developed by using the Picker PQ 2000 scanner and then applied to the volume data (19). Measurement, data entry, and volume calculation require, on average, 6 minutes per examination (14).

The DT (in days) in each patient was calculated by using the standard volumetric formula: DT = (t · ln 2)/[ln (Vf/Vi)], where t is the time (in days); Vf, the volume measured at final CT; and Vi, the volume measured at initial CT.

Data Analyses
For all analyses, the DT was converted to the linear function 365/DT, or reciprocal of DT (RDT). By using the RDT values measured from initial to final CT and assuming linear growth, we determined the time intervals needed for the tumor to increase in volume by growth thresholds of 5%, 10%, 15%, 20%, and 25% and compared these intervals with the current follow-up recommendations (15).

Growth detectability may be more difficult with smaller nodules: A higher threshold may be needed. To determine the distribution of volume growth thresholds necessary for growth detection in our study population, we used recently published data (21) on the repeatability of growth detection in simulated nodules to determine an exponential function, y = axb, where y is the percent volume change needed for detection and x is the tumor diameter. From these data, we derived the coefficients y = 61.18x–0.95. For each patient, we then calculated the diameter of a sphere with a volume equal to that of the initial tumor volume and substituted this diameter into the above formula (Fig 1).


Figure 1
View larger version (7K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1: Graph shows percent volume changes needed to detect growth of nodules with varying initial diameters, derived from the data of Kostis et al (21). Note that nodule size and volume change are inversely related—that is, small nodules require large volume changes.

 
Because the accuracy of growth rate measurement may be affected also by the time interval between CT examinations, we also performed an analysis to determine the RDT distribution by weighting each RDT according to the length of time (T) between the examinations. We first determined the median time interval (Tm) between the initial and final CT examinations for all patients; then we calculated a scale factor (S) for each patient (p): Sp = Tp/Tm. We then summed the products of Sp · RDTp for the 149 patients and determined a scaled median RDT and plotted the scaled RDT distribution. For various RDT intervals, we then plotted the scaled RDT distribution by summing these products for each patient within each interval.

Statistical Analyses
P < .05 was considered to indicate statistical significance. Statistical analyses were performed by using SAS, version 9.1 (SAS, Cary, NC), software. We compared histologic tumor type with RDT by using one-way analysis of variance. Age at diagnosis was compared with RDT, time interval between CT examinations, and initial tumor volume by using Pearson correlation. We also used Pearson correlation to compare RDT with initial tumor volume.

We divided patients into slow- and fast-growing tumor groups (according to median values) and evaluated the relationship between RDT and follow-up duration by using two-tailed t tests. The relationship between RDT and survival was analyzed by using the Kaplan-Meier method (with use of a binary model, where the RDT was either above or below the median) and a Cox proportional hazards model (with use of continuously distributed RDT values). June 2004 was the follow-up end point. All-cause mortality was used for analysis, as lung cancer–specific mortality proved difficult to determine because multiple comorbidities were common in our study population. We also compared the survival curves of patients with squamous cell carcinoma, adenocarcinoma, and bronchioloalveolar cell carcinoma by using a log-rank test. In addition, we investigated whether the RDT predicted survival among patients with adenocarcinoma, bronchioloalveolar carcinoma, and other tumor types by using a two-way analysis of variance model.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Ninety-six (64.4%) of the 149 tumors were in the upper lobes (Table 1). Malignancy was diagnosed (Table 2) by using tissue or fluid obtained at transthoracic fine-needle aspiration biopsy in 138 (92.6%) patients, at sputum analysis in five (3.4%), at transbronchial biopsy in three (2.0%), and at surgical excision in one (0.7%) patient. In two (1.3%) patients, histologic tissue was not obtained and the diagnosis was based on radiologic progression. According to greatest tumor diameter criteria, 99 (66.4%) patients had stage IA disease and 50 (33.6%) had stage IB disease at diagnosis.


View this table:
[in this window]
[in a new window]

  		Table 1. Locations of Lung Cancer in 149 Patients

 

View this table:
[in this window]
[in a new window]

  		Table 2. Histologic Tumor Type in 149 Patients

 
Median time interval between initial and final CT was 130 days (range, 25–2493 days). Mean section width was 5.5 mm. Tumor dimensions were measured on 2228 CT sections. Median tumor volume was 3000 mm3 (range, 22–103 189 mm3) at initial CT and 6213 mm3 (range, 261–499 658 mm3) at final CT (Fig 2). Median percent change in tumor volume (Fig 3) between initial and final CT was +54% (range, –81% to +10 759%). Twenty-one (14.1%) of the 149 tumors did not increase in volume between the initial and final CT examinations (Table 3).


Figure 2
View larger version (40K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2: Graph shows distribution of tumor volumes at initial and final CT examinations. Median initial tumor volume was 3000 mm3, while median final tumor volume was 6213 mm3.

 

Figure 3
View larger version (31K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3: Graph shows distribution of percent changes in tumor volume between initial and final CT examinations. The median percent change was +54%.

 

View this table:
[in this window]
[in a new window]

  		Table 3. Data for 21 Tumors That Did Not Show Growth

 
RDT Data
Median RDT between initial and final CT was 1.76, corresponding to a DT of 207 days. The RDT range was –7.24 to 14.18, corresponding to a DT range of –50 to 26 days. There was no significant difference in RDT (P > .05 for all comparisons) among the histologic types (Table 4, Fig 4). The scaled median RDT (Fig 5) was 1.43, or a DT of 255 days, and 5.7% of the scaled RDT distribution was less than 0.


View this table:
[in this window]
[in a new window]

  		Table 4. Tumor Growth Rate (RDT) from Initial to Final CT Examination, according to Histologic Type

 

Figure 4
View larger version (12K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4: Graph shows distribution of RDT values from initial to final CT examination among the 149 patients. Tumor volume reduction—that is, negative DT (dark gray bar), DT of greater than or equal to 1 year (light gray bar), and DT of less than 1 year (white bar) are illustrated.

 

Figure 5
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5: Graph shows scaled distribution of RDT values from initial to final CT examination among the 149 patients. Tp/130 = interscan interval for a particular patient/median interscan interval (130 days). The y-axis height is the sum of Tp/130 for all patients with an RDT within a particular range of values (eg, RDT = 0–1, 1–2, etc). Tumor volume reduction—that is, negative DT (dark gray bar), DT of greater than or equal to 1 year (light gray bar), and DT of less than 1 year (white bar) are illustrated. This scaling deemphasizes the values in patients with short time intervals between CT examinations.

 
There was no significant correlation between age at diagnosis and either interval between CT examinations (P = .16) or initial tumor volume (P = .43). However, age at diagnosis was negatively correlated with RDT (P = .047). Each 1-year increase in age at diagnosis was associated with a 6% decrease in RDT.

Volume Increase Thresholds
The plots of growth detectability over time derived by using different volume increase thresholds (Fig 6) show that the time at which 50% (n = 64) of the 128 growing tumors would have shown detectable growth, as extrapolated from the calculated RDTs, ranged from 2 weeks (5% threshold) to 8 weeks (25% threshold); 75% (n = 96) of the tumors would have shown detectable growth at 4 weeks (5% threshold) to 14 weeks (25% threshold); and 90% (n = 115) would have shown detectable growth at 8 weeks (5% threshold) and 37 weeks (25% threshold). With use of the current follow-up recommendations across the range of thresholds, detectable tumor growth would have occurred in 72%–95% of our study patients at 3 months, in 87%–98% at 6 months, in 95%–98% at 12 months, and in 98%–99% at 24 months after the baseline CT examination. Applying the data from Kostis et al (21) to our study population to calculate detectability thresholds revealed that 46 (35.9%) of the 128 growing nodules would have surpassed the volume growth detectability threshold within 1 week and 85 (66.4%) would have surpassed the threshold within 2 weeks (Fig 7).


Figure 6
View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6: Graph extrapolation of time (in weeks) needed to detect tumor growth, at various (5%–25%) thresholds of percent volume growth, among 128 growing tumors.

 

Figure 7
View larger version (13K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7: Graph shows time required for growth detectability in the study population, derived by using the formula of Kostis et al (21).

 
Slow- and Fast-growing Tumors
Forty-three (28.9%) of the 149 patients underwent surgery, while 106 (71.1%) did not; 33 (22.1%) patients elected not to undergo therapy. Twenty-five (34%) of 74 patients with slow- or not-growing tumors (Fig 8) and 18 (24%) of 75 with fast-growing tumors (Fig 9) underwent surgery (P = .21). No treatment was performed in 19 (26%) of the 74 patients with slow- or not-growing tumors or in 14 of the 75 (19%) patients with fast-growing tumors (P = .33).


Figure 8
View larger version (117K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8a: CT images in 73-year-old man with left upper lobe pleural-based lung nodule (arrow). Sections with the largest nodule cross section are shown. (a) At initial CT, the nodule volume is 1229 mm3. (b) At follow-up CT 10 months later, the volume is 1499 mm3 (DT, 1019 days). Fine-needle aspiration biopsy revealed bronchioloalveolar cell carcinoma.

 

Figure 8
View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8b: CT images in 73-year-old man with left upper lobe pleural-based lung nodule (arrow). Sections with the largest nodule cross section are shown. (a) At initial CT, the nodule volume is 1229 mm3. (b) At follow-up CT 10 months later, the volume is 1499 mm3 (DT, 1019 days). Fine-needle aspiration biopsy revealed bronchioloalveolar cell carcinoma.

 

Figure 9
View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 9a: CT images in 69-year-old man with left upper lobe lung nodule (arrow). Sections with the largest nodule cross section are shown. (a) At initial CT, the nodule volume is 22 mm3. (b) At follow-up CT 9 months later, the volume is 1535 mm3 (DT, 52 days). Fine-needle aspiration biopsy revealed squamous cell carcinoma.

 

Figure 9
View larger version (110K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 9b: CT images in 69-year-old man with left upper lobe lung nodule (arrow). Sections with the largest nodule cross section are shown. (a) At initial CT, the nodule volume is 22 mm3. (b) At follow-up CT 9 months later, the volume is 1535 mm3 (DT, 52 days). Fine-needle aspiration biopsy revealed squamous cell carcinoma.

 
Follow-up and Survival
The mean (± standard deviation) period of clinical follow-up after lung cancer diagnosis was 3.39 years ± 1.91 (median, 3.20 years). The mean follow-up period did not vary significantly between patients with slow- or not-growing tumors and those with fast-growing tumors (3.40 and 3.38 years, respectively; P = .95).

With use of the Kaplan-Meier model and the median RDT (1.76) as the cutoff value between the slow- or not-growing tumors and the fast-growing tumors, survival was significantly lower among the patients with fast-growing tumors (P = .002); Cox proportional hazards analysis revealed a hazard ratio for RDT of 1.1040 (P = .009). Among the patients with adenocarcinoma or bronchioloalveolar cell carcinoma, RDT did not enable the prediction of survival (mean RDTs for survivors vs nonsurvivors, 2.07 and 1.91, respectively). However, among the patients with other histologic types, RDT was significantly associated with survival (mean RDTs for survivors vs nonsurvivors, 3.14 and 1.46, respectively; P < .05). The area under the Kaplan-Meier survival curve for patients with squamous cell carcinoma was significantly smaller than that for patients with bronchioloalveolar cell carcinoma (P = .050) and nearly significantly smaller than that for the patients with adenocarcinoma (P = .06).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
ADVANCES IN KNOWLEDGE
 References
 
The median growth rate (DT, 207 days) of the stage I lung cancers in our study, which was measured volumetrically by using serial CT, is similar to previously reported radiographic and CT data (10,13,14). The range of growth rates was wider than that reported with radiography, and some tumors did not appear to grow, even after weighting for the interval between examinations. Older patients were more likely to have slow- or not-growing tumors, and decreased growth rate was associated with increased survival. Histologic type was not associated with growth rate.

Many investigators have estimated lung nodule growth by using chest radiography (7,9–11,22,23), and typically, almost all nodules that doubled in volume in less than 7 days or in more than 465 days were benign (24); the DT for most lung cancers was 1–18 months. However, with chest radiography volumetric measurement, tumors are assumed to be spherical. With CT, one can measure and sum the cross-sectional areas of each CT section to estimate volume. We observed a wider range of growth rates compared with the range reported from radiographic studies.

Previous CT studies (12–14) also have revealed a wider range of lung cancer DTs—from 32 days to essentially no growth—compared with chest radiography studies. Aoki et al (12) evaluated adenocarcinoma growth with CT, but they measured the tumors in only 10 patients, used mean tumor diameters to derive volumes, and did not differentiate CT from radiographic measurements; DTs ranged from 42 to 1486 days. By using CT, Hasegawa et al (13) measured the DTs (from tumor cross-sectional areas) of 61 lung cancers (not all stage I) and observed a DT range of 52–1733 days (mean, 452 days). We are unaware of previous study findings that show that lung cancers grow more slowly in older patients, but some believe that these tumors may behave differently in older patients (25).

An unexpected finding in our study was that 21 (14.1%) of the 149 patients had tumors that did not grow between the two CT examinations. When we weighted the growth rates for the time interval between examinations, only 5.7% of the scaled RDT distribution was less than 0 (indicating no growth). This decrease from 14.1% to 5.7% after scaling of RDT may have occurred because measurements based on short time intervals between examinations may be less reliable. The residual value (5.7%) may reflect not only error but also nonlinear tumor growth. A previous study of lung tumor growth measured with CT revealed no evidence of slowing lung tumor growth when the volume surpassed a threshold (14). However, there may be other causes of nonlinear growth, including variable immune response, necrosis and implosion, reexpansion of adjacent lung region with atelectasis, infection of the tumor, and variations in tumor angioneogenesis (14). In situ carcinomas may not grow (26); however, because nearly all of the tumors in our study population were diagnosed by using fine-needle aspiration biopsy, we cannot determine whether any were in situ cancers. In any event, our results show that a nodule that is stable or even decreasing in volume may not be benign. We also found that the patients with slow-growing tumors had longer survivals after diagnosis.

Our results indicate that the protocol most commonly used to follow up indeterminate nodules—follow-up CT at 3, 6, 12, and 24 months—may not be optimal. The basis for this schedule is obscure (16,17), and the distribution of growth detection across these follow-up time points is not known. However, by using different growth thresholds, we were able to apply our data to retrospectively predict when growth would be detectable among the tumors assessed in our study population. These thresholds depend on the nodule volume and the measurement technique. Small nodules and manual techniques require higher thresholds owing to the greater partial volume effects with small nodules and the greater intra- and interobserver variability with manual methods. With use of the highest threshold (25%), 50% of the tumors in our population would show growth at 8 weeks and 75% would show growth at 14 weeks. With use of the lowest threshold (5%), 50% would show growth at 2 weeks and 75% would show growth at 4 weeks.

Our data show that even if the highest threshold is used, follow-up CT of indeterminate nodules might be performed by using a compressed schedule, at half the time intervals currently used—1.5, 3, 6, and 12 months. With this schedule, one could identify approximately 95% of all growing lung tumors, do so more rapidly than is achieved in current practice, and institute therapy earlier. Moreover, as multidetector CT (which speeds acquisition and thus decreases motion artifacts and partial volume effects) and automated volume measurement software become more available (17), lung cancer growth may be seen in intervals as small as 4 weeks between CT examinations (27).

Among the 128 cancers that were growing in our study population, even the slowest growing tumors would not be detected at even 24-month follow-up (one of 128 patients at 5% threshold, two of 128 patients at 10%–25% thresholds). Biopsy was required to prove the malignancy of nodules that showed little or no growth. One disadvantage of a compressed follow-up schedule is that the nodules will tend to be smaller when growth is recognized. Radiologists may find an increased number of nodules that demonstrate growth at serial CT yet are too small for PET evaluation and fine-needle aspiration biopsy. This may result in more surgeries that yield noncancerous nodules.

A limitation of our study was that we did not measure biopsy-proved benign lesions, some of which may demonstrate growth. Another limitation was that manual measurements performed by one radiologist were used. However, Erasmus et al (28) observed improved consistency when one radiologist performed measurements on serial CT scans for the detection of non–small cell lung cancer. The compensatory equation that we applied to the measured tumor volumes to reduce partial volume effects may not be fully effective (19). Our measurement method was less precise than automated measurements, and we did not assess intra- or interobserver variability. The tumor volume measured with any method—including ours—may include adjacent atelectasis, pneumonitis, or scarring, and, thus, the apparent volume of the tumor may be increased. Necrosis or cavitation may decrease the measured volume.

We assumed that tumor growth was linear. Because this was a retrospective study, we could not control the time interval between the CT examinations, and short intervals may reduce the accuracy of growth rate measurement and thus skew the distribution of growth rates (a surplus of both very fast-growing and very slow-growing tumors). The mean tumor volume in our study population was larger than expected for indeterminate nodules that are followed up with serial CT; however, we found that the growth rate did not vary with initial tumor volume.

In conclusion, we determined that with serial CT tumor volume measurement, the median DT of stage I lung cancers is 207 days and there is wide variability in growth rates. Some biopsy-proved cancers decreased in volume between CT examinations. Slow growth was associated with longer survival and increased age at diagnosis. If smaller tumors grow at the same rate that the tumors did in our study, then using serial volumetric CT measurements may enable the current schedule for observing indeterminate nodules to be compressed. A prospective study is needed to determine the distribution of lung nodule growth detection across various follow-up time intervals.


   ADVANCES IN KNOWLEDGE
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
References
 


   FOOTNOTES
 

Abbreviations: DT = doubling time • RDT = reciprocal of DT

Authors stated no financial relationship to disclose.

Author contributions: Guarantor of integrity of entire study, S.G.J.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, S.G.J., H.T.W., M.T.; clinical studies, S.G.J., H.T.W.; statistical analysis, S.G.J., J.Y.; and manuscript editing, all authors


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 

  1. Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997;111(6):1710–1717.[Abstract/Free Full Text]
  2. Naruke T, Goya T, Tsuchiya R, et al. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988;96(3):440–447. [Published correction appears in J Thorac Cardiovasc Surg 1989;97(3):350.]
  3. Sone S, Takashime S, Li P, et al. Mass screening for lung cancer with mobile spiral computed tomography scanner. Lancet 1998;351:1242–1245.[CrossRef][Medline]
  4. Henschke CI, McCauley DI, Yankelevitz DF, et al. Early Lung Cancer Action Project: overall design and findings from baseline screening. Lancet 1999;354:99–105.[CrossRef][Medline]
  5. Kaneko M, Eguchi K, Ohmatsu H, et al. Peripheral lung cancer: screening and detection with low-dose spiral CT versus radiography. Radiology 1996;201:798–802.[Abstract/Free Full Text]
  6. Swensen SJ, Jett JR, Hartman TE, et al. CT screening for lung cancer: 5-year prospective experience. Radiology 2005;235:259–265.[Abstract/Free Full Text]
  7. Garland LH, Coulson W, Wollin E. The rate of growth and apparent duration of untreated primary bronchial carcinoma. Cancer 1963;16:694–701.[CrossRef][Medline]
  8. Spratt JS Jr, Spjut HJ, Roper CL. The frequency distribution of the rates of growth and the estimated duration of primary pulmonary carcinomas. Cancer 1963;16:687–693.[CrossRef][Medline]
  9. Steele JD, Buell P. Asymptomatic solitary pulmonary nodules, host survival, tumor size, and growth rate. J Thorac Cardiovasc Surg 1973;65:140–151.[Medline]
  10. Weiss W. Tumor doubling time and survival of men with bronchogenic carcinoma. Chest 1974;65:3–8.[Abstract/Free Full Text]
  11. Weiss W, Boucot KR, Cooper DA. Growth rate in the detection and prognosis of bronchogenic carcinoma. JAMA 1966;198:1246–1252.[CrossRef][Medline]
  12. Aoki T, Nakata H, Watanabe H, et al. Evolution of peripheral lung adenocarcinomas: CT findings correlated with histology and tumor doubling time. AJR Am J Roentgenol 2000;174:763–768.[Abstract/Free Full Text]
  13. Hasegawa M, Sone S, Takashima S, et al. Growth rate of small lung cancers detected on mass CT screening. Br J Radiol 2000;73:1252–1259.[Abstract]
  14. Winer-Muram HT, Jennings SG, Tarver RD, et al. Volumetric growth rate of stage I lung cancer prior to treatment: serial CT scanning. Radiology 2002;223:798–805.[Abstract/Free Full Text]
  15. Tan BB, Flaherty KR, Kazerooni EA, Iannettoni MD. The solitary pulmonary nodule. Chest 2003;123(1 suppl):89S–96S.
  16. Yankelevitz DF, Henschke CI. Does 2-year stability imply that pulmonary nodules are benign? AJR Am J Roentgenol 1997;168:325–328.[Free Full Text]
  17. Yankelevitz DF, Reeves AP, Kostis WJ, Zhao B, Henschke CL. Small pulmonary nodules: volumetrically determined growth rates based on CT evaluation. Radiology 2000;217:251–256.[Abstract/Free Full Text]
  18. Yankelevitz DF, Gupta R, Zhao B, Henschke CI. Small pulmonary nodules: evaluation with repeat CT—preliminary experience. Radiology 1999;212:561–566.[Abstract/Free Full Text]
  19. Winer-Muram HT, Jennings SG, Meyer CA, et al. Effect of varying CT section width on volumetric measurement of lung tumors and application of compensatory equations. Radiology 2003;229:184–194.[Abstract/Free Full Text]
  20. Jennings SG, Winer-Muram HT, Tarver RD, Farber MO. Lung tumor growth assessment with CT: comparison of diameter and cross-sectional area with volume measurements. Radiology 2004;231:866–871.[Abstract/Free Full Text]
  21. Kostis WJ, Yankelevitz DF, Reeves AP, Fluture SC, Henschke CI. Small pulmonary nodules: reproducibility of three-dimensional volumetric measurement and estimation of time to follow-up CT. Radiology 2004;231:446–452.[Abstract/Free Full Text]
  22. Garland LH. The rate of growth and natural duration of primary bronchial cancer. Am J Roentgenol Radium Ther Nucl Med 1966;96:604–611.[Medline]
  23. Weiss W. Peripheral measurable bronchogenic carcinoma: growth rate and period of risk after therapy. Am Rev Respir Dis 1971;103:198–208.[Medline]
  24. Nathan MH, Collins VP, Adams RA. Differentiation of benign and malignant pulmonary nodules by growth rate. Radiology 1962;79:221–232.[Medline]
  25. Filderman AE, Shaw C, Matthay RA. Lung cancer in the elderly. Clin Geriatr Med 1986;2:363–383.[Medline]
  26. Brambilla E, Travis WD, Colby TV, Corrin B, Shimosato Y. The new World Health Organization classification of lung tumours. Eur Respir J 2001;18:1059–1068.[Abstract/Free Full Text]
  27. Libby DM, Smith JP, Altorki NK, Pasmantier MW, Yankelevitz D, Henschke CI. Managing the small pulmonary nodule discovered by CT. Chest 2004;125:1522–1529.[Abstract/Free Full Text]
  28. Erasmus JJ, Gladish GW, Broemeling L, et al. Interobserver and intraobserver variability in measurement of non-small-cell carcinoma lung lesions: implications for assessment of tumor response. J Clin Oncol 2003;21:2574–2582.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Am. J. Roentgenol.Home page
J. G. Ravenel, P. Costello, and G. A. Silvestri
Screening for Lung Cancer
Am. J. Roentgenol., March 1, 2008; 190(3): 755 - 761.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2412051185v1
241/2/554    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jennings, S. G.
Right arrow Articles by Dowdeswell, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Jennings, S. G.
Right arrow Articles by Dowdeswell, I.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE