Original Research
Chest Imaging
July 2006

Software Volumetric Evaluation of Doubling Times for Differentiating Benign Versus Malignant Pulmonary Nodules

Abstract

OBJECTIVE. The purpose of our study was to evaluate the reliability of software-calculated doubling times for discerning malignant versus benign nodules.
MATERIALS AND METHODS. CT lung analysis volumetric software was used to retrospectively calculate the doubling times of 63 solid noncalcified nodules by comparing nodule volumes on baseline and follow-up CT scans obtained a median of 3.7 months apart. A final diagnosis based on validated criteria was available for all 63 nodules. All CT examinations were performed with 1.25-mm-thick slices on a four-detector unit. Taking 500 days as the upper value for malignancies, we evaluated whether the software-calculated doubling times could be used to distinguish malignant from benign solid nodules. We also examined whether the relative volume variation of benign nodules correlated with initial nodule size, interscan interval, or differences in contrast administration or exposure parameters between baseline and follow-up CT.
RESULTS. There were 52 benign and 11 malignant nodules. Benign nodules had a median doubling time of 947 days and a mean relative volume variation of -4.4% (range, -50% to 38%). Malignant nodules had a median doubling time of 117 days and a mean relative volume variation of 102% (22-462%). The sensitivity, specificity, and negative and positive predictive values of the volumetric software for diagnosing malignancy were 91% (95% confidence interval [CI], 0.59-1.00), 90% (95% CI, 0.79-0.97), 98% (95% CI, 0.89-1.00), and 67% (95% CI, 0.38-0.88), respectively. No correlation was found between the relative volume variation of benign nodules and their initial size, the interscan interval, or differences in contrast administration or exposure parameters between the two CT examinations.
CONCLUSION. Software-calculated pulmonary nodule doubling times of more than 500 days have a 98% negative predictive value for the diagnosis of solid malignant pulmonary nodules. This method may be useful for diagnosing malignant pulmonary nodules on follow-up CT.

Introduction

Pulmonary nodules are frequently found on chest CT. Their CT prevalence in the general population is not known precisely but can be estimated from lung cancer screening studies, in which pulmonary nodules were identified in 23-51% of subjects at baseline screening [1-3]. The use of MDCT technology, which allows thin slices to be acquired through the entire thorax, has increased the detection rate [4], especially for small nodules.
Most solid nodules identified incidentally or during screening are benign [1-3, 5], but this diagnosis is rarely evident on the basis of morphologic criteria alone. Although certain features on thin-section CT images can help to distinguish between small malignant and benign nodules [6], internal nodule characteristics show considerable overlap [7]. The presence of intranodular fat is one reliable sign of a benign lesion [7], but few nodules have that characteristic.
In theory, contrast-enhanced CT is applicable for the diagnosis of nodules larger than 5 mm, but its specificity is only 58% [8]. Fluorine-18 FDG PET is inaccurate for the diagnosis of nodules smaller than 10 mm in diameter [9, 10], which represent most nodules detected on CT [1-3, 5]. Because the probability of malignancy is relatively low in this size range [11], invasive management is not warranted. For all these reasons, nodules with a diameter of less than 10 mm are generally monitored by means of serial CT, with the aim of detecting a size increase suggestive of malignancy [12]. The maximal transverse diameter generally is used to monitor pulmonary nodules and to calculate their volume. However, volume changes estimated from 2D images may miss asymmetric growth [13]. Furthermore, manual measurements are not sufficiently reproducible to detect small variations in nodule diameter. The repeatability coefficient was 1.3 mm for the best observer in a study of intraobserver agreement for 2D CT measurements [14], suggesting that smaller variations cannot be considered clinically significant.
It is estimated that the doubling time of malignant tumors is generally between 30 and 500 days, with a mean of 100 days [15].
In this study, we evaluated software permitting automatic volume segmentation of pulmonary nodules and calculation of their doubling time. Using an upper cutoff of 500 days for the malignant doubling time, we calculated sensitivity, specificity, and positive and negative predictive values for the diagnosis of malignancy. We also analyzed the influence of factors such as initial nodule size, interscan interval, and differences between initial and follow-up CT examinations (contrast administration and exposure parameters) on the relative volume variation of benign noninfectious nodules.

Materials and Methods

This study was retrospective and did not affect patient management. It did not therefore require approval from our institutional review board or the patients' informed consent.

Nodule Selection and Imaging

Solid noncalcified nodules with a diameter smaller than 2 cm were selected for this study if they were detected on CT performed with a 1.25-mm collimation and a standard reconstruction algorithm and if a second CT scan obtained with these same two parameters was available, together with the final diagnosis. A keyword search on our department database for the years 2000-2003 yielded 63 nodules meeting these criteria. These 63 nodules had been found in 45 patients who ranged in age from 35 to 82 years (mean, 59 years) and had been examined with a CT unit (LightSpeed, GE Healthcare) for various clinical indications. The patients consisted of 15 women ranging in age from 41 to 82 years (median, 65 years) and 30 men ranging in age from 35 to 78 years (median, 54 years). The age distribution of the men and women was not significantly different (p = 0.08, Mann-Whitney test).
Twenty-seven of the 45 patients had a solitary nodule. The two largest nodules were selected in the 18 patients who had more than one nodule. CT had been performed for follow-up of thoracic or extrathoracic malignancy in 17 patients and to characterize a pulmonary nodule that had been discovered incidentally on chest radiography in 15 patients. In the remaining 13 patients, the nodules had been found incidentally on CT scans acquired for suspected pulmonary embolism in seven cases; hemoptysis in three cases; and pulmonary fibrosis, cardiomyopathy, and pleuro-pneumonia in one case each.
The initial CT acquisitions were performed with (20 cases) or without (43 cases) contrast medium, depending on the indication.
The CT acquisition that served to calculate the nodule doubling time was generally performed 3 months after the first; in our institution, this CT follow-up time is standard for patients with indeterminate nodules measuring between 5 and 10 mm. For patients with nodules smaller than 5 mm in diameter, the interval was between 6 and 12 months. Nodules with evidence of malignancy on the baseline examination were reevaluated 3-8 weeks later, at the time of core biopsy or surgical resection.

Software Description and Doubling Time Calculation

The nodule volume doubling time was calculated on a workstation (Advantage Workstation 4.0, GE Healthcare Europe) running software (CT Advanced Lung Analysis, GE Healthcare Europe). This software segments pulmonary nodules with a combination of watershed segmentation and shape-analysis techniques. It is dedicated to the evaluation of small lesions, not large masses. Different shape-analysis algorithms are used depending on the type of nodule, which is recognized automatically—that is, juxtavascular (strong vascular connections); juxtapleural (nodule connected to the pleural surface); and well circumscribed (few vascular connections and no pleural contact). CT acquisitions with a slice thickness of no more than 1.25 mm and a standard reconstruction algorithm are recommended for accurate segmentation. The software compares nodule volumes between baseline and follow-up CT and calculates the doubling time if growth is detected. The same CT parameters (field of view, exposure parameters) should be used for volume comparison and accurate doubling-time estimation. The relative volume variation (RVV) is obtained with the following equation:
\[RVV=V_{f}-V_{i}{/}V_{i},\]
where Vf is the final nodule volume and Vi, the initial volume. The nodule doubling time cannot be calculated when the relative volume variation is negative.
On the basis of the literature [15, 16], we assumed that a doubling time of less than 500 days corresponded to a malignant lesion for solid nodules. Using this cutoff value, we calculated the sensitivity, specificity, and positive and negative predictive values of the software for diagnosing malignancy.
Noninfectious benign nodules normally remain stable over time, and their relative volume variation between two consecutive CT scans should therefore be close to zero. Two of the 52 benign nodules were found to be infectious, and both fully resolved on late control examinations. These two nodules were thus excluded from further analysis. For the remaining 50 benign nodules, we analyzed the influence on the relative volume variation of the following factors: initial nodule size, nodule type (juxtapleural, juxtavascular, or well circumscribed), interscan interval, and differences in the use of contrast material and in the exposure parameters between the two CT examinations.

Final Diagnosis

The final diagnosis of malignancy for the 11 malignant lesions was based on pathologic examination of core biopsies, surgical specimens, or both. Other lesions were confirmed to be benign if they did not grow for at least 2 years on CT (42 nodules), showed no 18F-FDG uptake and measured at least 10 mm (two nodules), or had morphologic features typical of a benign lesion (seven nodules contained fat). The last benign nodule in this series was confirmed to be benign at surgery.

Statistical Analysis

Statistical calculations were performed using statistics software (SAS version 8.2, SAS Institute). The performance of the CT Advanced Lung Analysis software was assessed in terms of its sensitivity, specificity, and positive and negative predictive values for malignancy; the associated 95% confidence intervals (CIs) were calculated with the exact binomial method. Age distribution was compared between men and women using the Mann-Whitney test.
Because the relative volume of variation was normally distributed, the correlation between relative volume variation and the studied parameters was tested with parametric methods—namely, Fisher's exact test (analysis of variance or covariance) for quantitative variables and qualitative variables with more than two techniques and the Student's t test for qualitative variables with two categories.
Regarding the absolute value of the relative volume of variation (AVRVV), we considered two groups of benign nodules: group A comprised benign nodules for which AVRVV was less than 10%, and group B comprised benign nodules with 10% or higher AVRVV values. The effects of potential factors of variability were compared in these two groups of nodules using the Student's t test for normally distributed quantitative variables such as the initial nodule size, the Mann-Whitney test for other quantitative variables such as the interscan interval, and the chi-square test for categoric variables such as the nodule type.
Fig. 1A Malignant pulmonary nodule in 54-year-old woman who is not a smoker. Unenhanced axial transverse CT image shows 9-mm round nodule with smooth margins in right upper lobe. Patient had history of ovarian malignancy. No prior CT examination was available.
Fig. 1B Malignant pulmonary nodule in 54-year-old woman who is not a smoker. On second CT examination, 2 months after A, size of nodule has not changed on visual estimation. Measured diameter variation is only 0.9 mm, which is a nonsignificant change.
Fig. 1C Malignant pulmonary nodule in 54-year-old woman who is not a smoker. Volume of nodule on follow-up CT, 2 months after A, is 547 mm3. Relative volume variation is thus 41%, and calculated doubling time is 117 days, suggesting malignancy. Nodule was surgically removed and was found to be pulmonary metastasis expressing same hormone receptors as primary ovarian tumor.

Results

The nodules selected on baseline CT acquisitions measured 3-19 mm (mean, 8.9 ± 3.8 [SD] mm). Of the 63 nodules, six nodules (9.5%) measured less than 5 mm; 32 (51%), 5 to < 10 mm; 19 (30%), 10 to < 15 mm; and six (9.5%), 15-19 mm. Twenty-seven nodules (43%) were well circumscribed, 16 (25%) were juxtapleural, and 20 (32%) were juxtavascular. The median interscan interval was 3.7 months (range, 0.8-12.5 months). Eleven nodules (17%) were malignant on pathologic examination, whereas the remaining 52 nodules (83%) were shown to be benign.

Malignant Nodules

Seven of the 11 malignant nodules corresponded to primary lung carcinomas and four to metastases. The interscan interval ranged from 0.8 to 6 months (median, 1.9 months). The relative volume variation of the malignant lesions ranged from 22% to 462% (mean, 102%; median, 55%). The diameter variation, measured with electronic calipers on a PACS screen, was more than 2 mm for six of the 11 nodules and less than 1 mm (not significant [NS]) for the other five nodules (Figs. 1A, 1B, 1C, 2A, 2B, and 2C). These five nodules were rescanned within 8 weeks after the baseline CT examination, at the time of core biopsy or surgical resection. On this basis, the sensitivity of the software-calculated doubling time for malignancy was 91% (95% CI, 0.59-1.00), whereas the sensitivity of manual diameter-change measurement was 54% (95% CI, 0.23-0.83).
The software-calculated doubling times of the malignant nodules were always less than 500 days (range, 37-297 days; mean, 116 days; median, 91 days), except for one adenocarcinoma (646 days) (Figs. 3A, 3B, and 3C). On this basis, the sensitivity of the software-calculated doubling time for malignancy was 91% (95% CI, 0.59-1.00). The mean and median doubling times for all 11 malignant lesions were 164 and 117 days, respectively.

Benign Nodules

Twenty-three of the 52 benign lesions had a positive relative volume variation, allowing the doubling time to be calculated. Benign nodules had a median doubling time of 947 days.
The doubling time was less than 500 days in five false-positive cases (range, 138-396 days). Thus, the specificity of the software-calculated doubling time was 90% (95% CI, 0.79-0.97). Two of these cases were due to erroneous automated nodule segmentation because of pleural or mediastinal connections that were evident on visual analysis. The remaining three cases were due to partial volume averaging at the top of the nodule modifying the segmentation from the first CT examination to the second. Except for two infectious nodules that shrank, the other 50 benign nodules remained stable over time (Figs. 3A, 3B, 3C, 4A, and 4B) and their relative volume variation did not correspond to a real volume change. Twenty-nine of these 50 nodules had a difference in tube current (mAs) exposure of more than 50 mAs between the two scans. In 15 cases, only one scan involved contrast administration.
Fig. 2A Malignant pulmonary nodule in 65-year-old man who is a smoker. Unenhanced axial transverse CT image shows 11-mm lobulated nodule in right upper lobe, which is likely to be malignant given patient's age and smoking status and size and lobulated contours of nodule. Nodule was detected on routine chest radiograph with patient being asymptomatic.
Fig. 2B Malignant pulmonary nodule in 65-year-old man who is a smoker. No obvious change in nodule size is apparent on second CT examination, which was performed preoperatively. Diameter variation measured with manually positioned electronic calipers is 0.8 mm.
Fig. 2C Malignant pulmonary nodule in 65-year-old man who is a smoker. Software-measured volume on second CT examination is 993 mm3, representing 29% increase. Software-estimated doubling time is 296 days, suggesting malignancy. Pathologic examination after surgical resection showed stage I poorly differentiated large cell carcinoma.
The relative volume variation values of these 50 benign nodules ranged from -50% to 38% (mean, -4.4%) and were normally distributed. No correlation was found between the relative volume variation and the initial nodule size (p = 0.84, F test, NS) (Fig. 5A), the interscan interval (p = 0.59, F test, NS), the nodule type (p = 0.20, F test, NS) (Fig. 6), differences in contrast administration (p = 0.12; Student's t test, NS), or differences in tube current settings (p = 0.58, Student's t test, NS).
The absolute value of the relative volume variation, AVRVV, ranged between 0% and 50% (mean, 13.5%; median, 9%). The proportions of benign nodules with an AVRVV below 10% (group A) and an AVRVV of at least 10% (group B) were 54% (27/50) and 46% (23/50), respectively. The mean nodule size was 7.9 mm (± 2.8 [SD] mm) and 8.1 mm (± 3.4 mm) in benign groups A and B, respectively (p = 0.82, Student's t test, NS) (Fig. 5B). The median interscan interval was 3.7 months in group A and 3.5 months in group B (p = 0.99, Mann-Whitney test, NS). The proportions of well-circumscribed, juxtapleural, and juxtavascular nodules were, respectively, 33% (9/27), 37% (10/27), and 30% (8/27) in group A and 57% (13/23), 17% (4/23), and 26% (6/23) in group B (p = 0.19, chi-square test, NS). The proportions of cases with and without differences in contrast administration were 78% (21/27) and 22% (6/27), respectively, in group A and 61% (14/23) and 39% (9/23) in group B (p = 0.19, chi-square test, NS). The proportions of cases with small and large differences in tube current exposure were 44% (12/27) and 56% (15/27), respectively, in group A, and 39% (9/23) and 61% (14/23) in group B (group A vs B, p = 0.70, chi-square test, NS) (Fig. 7). These results are summarized in Table 1.
TABLE 1: Parameters Tested for Their Influence on the Relative Volume Variation (RVV) of Benign Nodules
Absolute Value of the RVV (AVRVV)p
Parameter< 10%≥ 10%AVRVVaRVV
No. of nodules2723  
Initial nodule size (mm) (mean ± SD)7.9 ± 2.758.1 ± 3.360.820.84
Interscan interval (month) (median)3.7 [1.7, 5.0]3.5 [1.7, 6.1]0.990.59
No. (%) of nodules    
   Nodule type  0.190.20
      Juxtapleural10 (37)4 (17)  
      Juxtavascular8 (30)6 (26)  
      Well circumscribed9 (33)13 (57)  
Contrast administration  0.190.12
   Similar between the two scans21 (78)14 (61)  
   Different between the two scans6 (22)9 (39)  
Exposure parameters  0.700.58
   < 50 mAs difference beween the two scans12 (44)9 (39)  
   ≥ 50 mAs difference between the two scans
15 (56)
14 (61)


Note—Numbers in brackets are first and third quartiles (Q1, Q3); numbers in parentheses are percentages
a
Considering two groups: AVRVV < 10% vs ≥ 10%
Fig. 3A Benign nodule in 50-year-old man who is a smoker. Unenhanced CT scan obtained to confirm chest radiographic findings shows 10-mm noncalcified nodule in right upper lobe. Centrilobular emphysema is also present.
Fig. 3B Benign nodule in 50-year-old man who is a smoker. Three-dimensional surface-rendering view of nodule segmentation on first CT follow-up. Volume variation at 6 weeks is only 1%, giving estimated doubling time of more than 11 years.
Fig. 3C Benign nodule in 50-year-old man who is a smoker. Nodule segmentation on CT scan obtained 6 months after A. Volume variation remains very small (2%), and software-estimated doubling time is more than 17 years. This nodule had remained unchanged at 3 years.
Fig. 4A Benign nodule in 44-year-old man who is a smoker. Unenhanced CT axial transverse image shows 6-mm perifissural opacity. This nodule is likely to be intrapulmonary lymph node.
Fig. 4B Benign nodule in 44-year-old man who is a smoker. Segmentation of nodule on CT scan obtained 10 months after A. Three-dimensional surface-rendering view of segmented nodule shows its triangular shape. Nodule morphology and shape are better analyzed on volumetric than on 2D CT images. This nodule is likely to be intrapulmonary lymph node. Relative volume variation is 6%, and estimated nodule doubling time is 593 days. Nodule remained unchanged at 28 months.
Fig. 5A Volume variation of benign nodules according to their initial size. Plot of relative volume variation against initial nodule size. No correlation was found (p = 0.84, F test).
Fig. 5B Volume variation of benign nodules according to their initial size. Nodule size distribution in benign group A (absolute volume variation, < 10%) and benign group B (absolute volume variation, ≥ 10%). Mean nodule size and nodule size distribution were similar in groups A and B (p = 0.82, Student's t test).
Fig. 6 Volume variation of benign nodules according to nodule type. Plot of relative volume variation against nodule type. No correlation was found (p = 0.2, F test).
Fig. 7 Volume variation of benign nodules according to differences in tube current (mAs) exposure between two successive CT scans. Relative volume variation of nodules scanned with less than 50-mAs difference and nodules scanned with more than 50-mAs difference between two CT scans. Distribution and mean relative volume variation were not different between two groups.

Discussion

Indeterminate pulmonary nodules that measure less than 10 mm in diameter represent most nodules found on CT [1-3]. They are generally investigated by means of CT follow-up [12], with growth being suggestive of malignancy. However, manual measurement has been shown to be unreliable for detecting the growth of such small nodules [14].
Automated volumetric measurement methods are more reproducible [17], a key factor in serial comparisons, than manual measurements. However, volume measurements may be influenced by the level of inspiration or the patient's position and by differences in successive CT protocols, even though protocols are, in principle, standardized. In our institution, the CT protocol for nodule follow-up consists of 1.25-mm-thick helical acquisitions performed without contrast administration, with the exposure parameters depending on the patient's weight (70 mAs for a 70-kg patient). However, only 21 of the 50 benign nodules in this study were examined with a less than a 50-mAs difference between the two scans, and contrast administration status was identical between the two scans in only 35 cases.
Here, we evaluated the capacity of a volumetric software program to distinguish between malignant and benign solid nodules on the basis of their doubling times. Assuming a malignant doubling time cutoff of 500 days, we obtained a sensitivity and specificity of 91% and 90%, respectively. This time of 500 days is generally accepted as the upper limit of the doubling time for malignant pulmonary lesions, even though some such lesions grow more slowly (especially some adenocarcinomas in elderly patients) [18]. One nodule in our study, corresponding to an adenocarcinoma, had a software-calculated doubling time of 646 days. The use of a higher malignant doubling time would have increased the number of false-positives in our study. With a 700-day cutoff for malignant doubling, there would have been no false-negatives but three additional false-positives, reducing the specificity from 90% to 85%. The mean doubling time for the 11 malignant cases in our study was 164 days, a value close to the 149 days reported by Hasegawa et al. [16] for malignant solid nodules. Our series did not include any partially solid nodules or nodules with ground-glass opacities (likely bronchioloalveolar carcinomas), which are slow growers, with reported doubling times of more than 1,000 days [16].
The manually measured diameter variation was less than 1 mm (i.e., below the manual measurement error) for five of the 11 malignant nodules identified here. These lesions would have been left to follow-up rather than diagnosed as malignancies if they had been evaluated only on the basis of their interscan diameter change within 8 weeks after baseline CT. This underlines the need to use another measurement method, such as volumetry, for at least short interscan intervals.
The median interval between baseline and follow-up CT was less than 2 months for the 11 malignant nodules in this study. This suggests that the recall time for the first follow-up CT may be shortened when volumetric software is used.
We found no correlation between the relative volume variation of benign nodules and any of the potential factors of variability we tested, such as differences in tube current exposure or contrast administration between CT protocols, the interscan interval, the nodule type, or the nodule size on baseline CT. The relative volume variation was expected to correlate with initial nodule size in our evaluation, particularly because Yankelevitz et al. [13] have found that the relative error in volumetric measurements of nodule phantoms depends on nodule size. However, when evaluated on real nodules, this is not observed. In the study by Revel et al. [17], in which three radiologists performed three consecutive measurements of the nodules' volumes on the basis of a single CT acquisition, the software repeatability was not correlated with the nodules' sizes.
This was also observed by Wormanns et al. [19], on the basis of two consecutive CT acquisitions performed at a 10-minute interval in patients with pulmonary metastases. That study comprised 151 nodules with a mean diameter of 7.4 mm. The volume measurement error between the two CT scans did not correlate with the mean nodule volume in the study of Wormanns et al.
Because the relative volume variation of benign nodules is not influenced in our study by nodule characteristics or differences between the successive CT protocols, the main sources of variation appear to be differences in patient positioning or in the degree of inspiration. Because the software segmentation process is partly based on a density threshold, it may be affected by factors modifying the density of the nodule environment, such as the amount of air around the nodule. It would be useful if subsequent versions of volumetric software indicated the variation in whole-lung volume between the two CT acquisitions used for doubling-time calculation.
This study has several limitations. One is the relatively small number of nodules evaluated (n = 63). There are two main reasons for that: One is that diagnostic validation of benign lesions requires at least 2 years of stability on CT follow-up. We could not include nodules with a shorter CT follow-up. The second reason is that we evaluated no more than two nodules per patient to avoid a potential patient-effect bias.
Another limitation is the relatively small proportion of malignant lesions in our series, in which only 11 (17%) of 63 nodules were malignant. However, this proportion reflects that seen in daily practice. Among the large number of nodules found on CT, most are benign. Furthermore, most malignant nodules are identified during initial imaging studies and (fortunately) are not referred for follow-up. The nodules in our series were selected if they were solid, non-calcified, less than 2 cm and if two CT scans obtained with a 1.25-mm collimation and a standard reconstruction algorithm were available, together with the final diagnosis. We believe this is the right approach to obtain a nonbiased series. A series with a majority of malignant lesions would have been strongly biased and could not have been used to estimate the specificity of the test.
The final diagnosis of the benign or malignant nature of the nodules was based on histology only for the 11 malignant lesions and for one benign nodule. For the remaining benign nodules, diagnosis was mainly based on their long-term—2 years or more—stability (42 nodules). This has the potential risk of misdiagnosing malignancy in some slowly-growing lung cancers, and thus of overestimating the sensitivity of the doubling time in our study. However, slow growth mainly concerns non-solid lesions that were not included in our series, which evaluated only solid nodules.
The imaging examinations were performed on a 4-MDCT scanner, which is no longer state-of-the-art. However, this technology allows one to acquire 1.25-mm-thick slices, which are considered adequate for accurate segmentation with the software we used. It would be of interest to determine whether nodule segmentation is better with 0.625-mm-thick slices. Partial volume effects on nodule borders should be less marked, especially for small nodules; thus, the segmentation should be improved.
Finally, the retrospective nature of the study means that we are unable to propose an optimal recall time for evaluating the growth of lung nodules. One theoretic advantage of using dedicated volumetric software would be to detect eventual growth earlier. The median interscan interval for the malignant lesions in our series was only 1.9 months, which is shorter than the traditional 3-month interval used for first CT follow-up, on the basis of recommendations in lung cancer screening studies. Patient management was not modified in this series because the software was used retrospectively. However, our results have led us to reevaluate high-risk patients with nodules close to 1 cm in diameter earlier—at 8 weeks instead of 3 months. Nodules stable at 3 months are reevaluated only at 1 and 2 years; the 6-month manual diameter follow-up is no longer performed. Prospective studies are needed to determine the optimal first and subsequent recall times for software volumetric measurements.
In conclusion, this study shows that software-calculated doubling time, estimated on the basis of a second CT study performed a median of 3.7 months after the first, has a negative predictive value of 98% for malignancy in small, solid, initially indeterminate pulmonary nodules. Prospective studies are needed to determine the optimal recall time when using dedicated volumetric software.

Footnote

Address correspondence to M.-P. Revel ([email protected]).

References

1.
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
2.
Diederich S, Wormanns D, Semik M, et al. Screening for early lung cancer with low-dose spiral CT: prevalence in 817 asymptomatic smokers. Radiology 2002; 222:773-781
3.
Swensen SJ, Jett JR, Sloan JA, et al. Screening for lung cancer with low-dose spiral computed tomography. Am J Respir Crit Care Med 2002; 165:508-513
4.
Fischbach F, Knollmann F, Griesshaber V, Freund T, Akkol E, Felix R. Detection of pulmonary nodules by multislice computed tomography: improved detection rate with reduced slice thickness. Eur Radiol 2003; 13:2378-2383
5.
Sone S, Li F, Yang ZG, et al. Results of three-year mass screening programme for lung cancer using mobile low-dose spiral computed tomography scanner. Br J Cancer 2001; 84:25-32
6.
Li F, Sone S, Abe H, Macmahon H, Doi K. Malignant versus benign nodules at CT screening for lung cancer: comparison of thin-section CT findings. Radiology 2004; 233:793-798
7.
Erasmus JJ, Connolly JE, McAdams HP, Roggli VL. Solitary pulmonary nodules: Part I. Morphologic evaluation for differentiation of benign and malignant lesions. RadioGraphics 2000; 20:43-58
8.
Swensen SJ, Viggiano RW, Midthun DE, et al. Lung nodule enhancement at CT: multicenter study. Radiology 2000; 214:73-80
9.
Detterbeck FC, Falen S, Rivera MP, Halle JS, Socinski MA. Seeking a home for a PET: part 1. Defining the appropriate place for positron emission tomography imaging in the diagnosis of pulmonary nodules or masses. Chest 2004; 125:2294-2299
10.
Nomori H, Watanabe K, Ohtsuka T, Naruke T, Suemasu K, Uno K. Evaluation of F-18 fluorodeoxyglucose (FDG) PET scanning for pulmonary nodules less than 3 cm in diameter, with special reference to the CT images. Lung Cancer 2004; 45:19-27
11.
Henschke CI, Yankelevitz DF, Naidich DP, et al. CT screening for lung cancer: suspiciousness of nodules according to size on baseline scans. Radiology 2004; 231:164-168
12.
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
13.
Yankelevitz DF, Reeves AP, Kostis WJ, Zhao B, Henschke CI. Small pulmonary nodules: volumetrically determined growth rates based on CT evaluation. Radiology 2000; 217:251-256
14.
Revel MP, Bissery A, Bienvenu M, Aycard L, Lefort C, Frija G. Are two-dimensional CT measurements of small noncalcified pulmonary nodules reliable? Radiology 2004; 231:453-458
15.
Geddes DM. The natural history of lung cancer: a review based on rates of tumour growth. Br J Dis Chest 1979; 73:1-17
16.
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
17.
Revel MP, Lefort C, Bissery A, et al. Pulmonary nodules: preliminary experience with three-dimensional evaluation. Radiology 2004; 231:459-466
18.
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
19.
Wormanns D, Kohl G, Klotz E, et al. Volumetric measurements of pulmonary nodules at multi-row detector CT: in vivo reproducibility. Eur Radiol 2004; 14:86-92

Information & Authors

Information

Published In

American Journal of Roentgenology
Pages: 135 - 142
PubMed: 16794167

History

Submitted: July 15, 2005
Accepted: October 10, 2005
First published: November 23, 2012

Keywords

  1. chest
  2. comparative studies
  3. computer applications—3D
  4. CT
  5. lung
  6. lung cancer
  7. MDCT
  8. oncologic imaging

Authors

Affiliations

Marie-Pierre Revel
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Département de Radiologie, Hôpital Européen Georges Pompidou, Paris, F-75015 France.
Aurelie Merlin
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Département de Radiologie, Hôpital Européen Georges Pompidou, Paris, F-75015 France.
Severine Peyrard
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Centre d'Investigations Cliniques, Hôpital Européen George Pompidou, Paris, F-75015 France.
Rached Triki
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Département de Radiologie, Hôpital Européen Georges Pompidou, Paris, F-75015 France.
Sophie Couchon
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Département de Radiologie, Hôpital Européen Georges Pompidou, Paris, F-75015 France.
Gilles Chatellier
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Unité de Recherché Clinique, Hôpital Européen George Pompidou, Paris, F-75015 France.
Guy Frija
Département de Radiologie, Assistance Publique des Hôpitaux de Paris, 20 rue Leblanc, Paris, F-75015 France.
Université Paris-Descartes, Faculté de Médecine, Paris, F-75015 France.
Département de Radiologie, Hôpital Européen Georges Pompidou, Paris, F-75015 France.

Metrics & Citations

Metrics

Citations

Export Citations

To download the citation to this article, select your reference manager software.

Articles citing this article

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media