2239

Tumor Doubling Time and Prognostic

Assessment of Patients with Primary
Lung Cancer
Katsuo Usuda, M.D., Yasuki Saito, M.D., Motoyasu Sagawa, M.D.,
Masami Sato, M.D., Keiji Kanma, M.D., Satomi Takahashi, M.D., Chiaki Endo, M.D.,
Yan Chen, M.D., Akira Sakurada, M.D., and Shigefumi Fujimura, M.D.

Background. Relationships between tumor doubling

time (DT) and other prognostic factors and the risk of
death related to these factors are not yet fully understood.
Methods. Tumor doubling time of primary lung carcinomas of 174 patients, detected in a limited number of
local municipalities during a limited period, was calculated using the Schwartz formula. Survival rate of the 174
patients was compared with reference to categories of
prognostic factors (univariate analyses) and significant
factors affecting survival were identified by multivariate
analyses using the Cox proportional hazard model.
Results. Tumor doubling time had a log normal distribution. There was a significant difference in mean DT
in relation to sex, smoking history, presence of symptoms, cell type, primary tumor factor, and stage. Univariate analyses showed a significant difference in survival
in relation to DT, age, sex, method of tumor detection,
smoking history, symptoms, therapy, cell type, primary
tumor (T) factor, regional lymph node (N) factor, distant
metastasis (M) factor, and stage. Multivariate analyses
using the Cox's proportional hazard model in a stepwise
fashion identified a final set of five significant variables:
N factor (P = 0.0001); therapy (P = 0.0016);M factor (P =
0.0017);T factor ( P = 0.0018), and DT ( P = 0.0152).
Conclusions. Tumor doubling time was an independent and significant prognostic factor for lung cancer patients. Cancer 1994;742239-44.
Key words: primary lung cancer, growth rate, tumor doubling time, prognostic factor, survival rate, multivariate
analysis, Cox proportional hazard model.

From the Department of Thoracic Surgery, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan.
Supported in part by a grant-in-aid for cancer research from the
Japanese Ministry of Health and Welfare (2-4).
The authors thank M. Motomiya for his kind criticism.
Address for reprints: Katsuo Usuda, M.D., the Department of
Thoracic Surgery, Institute of Development, Aging and Cancer, Tohoku University, 4-1 Seiryo-Machi, Aoba-Ku, Sendai 980, Japan.
Received January 7, 1994; revisions received April 11, 1994, and
June 20,1994; accepted June 20,1994.

The growth rate of cancer, which reflects the degree of

malignancy, is closely related to prognosis. There have
been numerous reports on the growth rate of primary
lung cancer.'-9 However, relationships between the
growth rate of cancer and other prognostic factors are
not fully understood. In the current study, tumor doubling time (DT) was calculated for primary lung cancers
that were detected in a limited number of local municipalities during a limited period. Survival rate was compared in relation to the categories of prognostic factors;
in addition, multivariate analyses were used to evaluate
the risk of death related to DT and other prognostic factors.

Patients and Methods

The subjects of the current study were 174 patients (123
men and 51 women) in 46 local municipalities during
the period from January 1985 to December 1986. These
were the only patients who had chest X-rays taken between 3-12 months before detection of lung cancer.
There was no selection. Seven patients with a central
tumor were excluded because of postatelectatic changes
that precluded measurements. All of the patients in our
study had a tumor shadow that was easier to detect in
the peripheral lung field. The age of patients ranged
from 33 to 86 years (mean, 67 years). Of the 174 patients, 129 had disease detected by the annual lung cancer screening trial in Miyagi Prefecture. Forty-five had
disease detected in the Hospital attached to the Institute
of Development, Aging and Cancer, Tohoku University, Sendai Kousei Hospital, or Miyagi Prefectural
Semine Hospital, where they reported symptoms or
were treated for other diseases. One hundred twentysix patients with lung cancer underwent resection, and
48 underwent nonsurgcal treatment. Thirty-three received chemotherapy or radiotherapy. No treatment
was given to 15 patients. There were 86 adenocarcinomas, 67 squamous cell carcinomas, 7 small cell carcino-

2240

CANCER October 15,2994, Volume 74, No. 8

shadow of tumor

Figure 1. Calculation of tumor doubling time (DT). t: time between

the initial and second measurement; V,: tumor volume at the initial
measurement; a,: maximum dimension of tumor at the initial
measurement; b,: perpendicular dimension of tumor that crosses a,
at the midpoint; V,: tumor volume at the second measurement;
a,: maximum dimension of tumor at the second measurement;
b,: perpendicular dimension of tumor that crosses a, at the midpoint.

mas, 12 large cell carcinomas, 1 carcinoid, and 1adenosquamous carcinoma. Of the 126 patients with resected
lung cancer, 2 had a Stage I localized bronchioloalveolar carcinoma and 1 had a Stage I11 bronchioloalveolar carcinoma. In 48 patients who underwent nonsurgical treatment, diagnosis was made by cytology,
and the number of patients with a localized bronchioloalveolar carcinoma could not be confirmed.
A chest X-ray film taken at least 3 months before
the time when a tumor was detected was available for
all patients. The volume DT was calculated using the
Schwartz formula," as shown in Figure 1. Tumor size
was measured separately by two radiologists on posteroanterior roentgenograms. Interobserver difference
was less than 4 mm for a well defined tumor shadow
and 7 mm at the most for an ill defined tumor shadow.
DT was calculated based on the mean obtained by the
two observers. Intraobserver difference was less than 3
mm. Measurements were taken solely from the X-rays.
There was a strong correlation between the size of tumor on X-rays and the size of tumor measured by a pathologist after excision. The correlation coefficient was
0.9573 (P < 0.0001) in 69 patients who underwent resection and in whom a pathologic size of tumor was
recorded.
The minimum size of lung cancer that can be detected on a chest X-ray film has been reported to be 6
mm.1.3.11 Of the 174 patients, 136 had a tumor shadow
on a chest X-ray taken within 3-12 months before the
first detection of a lung cancer. Thirty-eight patients
had no tumor shadow on a chest X-ray taken within 312 months before the first detection of a lung cancer,
and an assumption was made that a tumor was there
and that its greatest dimension was 6 mm (being smaller
than the limit of detection) at the time when the reference film was taken. In the 38 patients, 34 had disease
detected by the annual screening trial in Miyagi Prefecture, and the last normal chest X-rays were taken in the
previous year. The mean duration of time between the
last normal chest X-ray and the measured abnormal
chest X-ray in the patients with disease detected by
screening was 12 months. Four other patients had dis-

ease detected by hospital X-rays, and the mean duration

of time between the last normal chest X-ray and the
measured abnormal chest X-ray in the patients with disease detected by hospital X-rays was 11 months. There
was no difference in the mean time interval between the
last normal chest X-ray and the abnormal chest X-ray in
the patients with disease detected by screening as opposed to the patients with disease detected by hospital
X-rays.
As of June 1991, 64 of the 174 patients were alive,
99 were dead of lung cancer, and 11 were dead of diseases other than cancer.
Survival rates were compared in relation to categories of prognostic factors (univariate analyses), and an
attempt was made to evaluate the risk of death related
to DT and other prognostic factors by multivariate analyses using the Cox proportional hazard model."
Tumor histologic type and stage were documented
based on the criteria of clinical and pathologic records
of lung cancer of the Japan Lung Cancer Society,13
which essentially is the same as that of the International
Union Against Cancer (UICC) clas~ification.'~
Data of
measurement were documented in terms of arithmetic
mean k standard deviation (AM k SD) or in terms of
geometric mean (GM). The significance of difference in
mean value was checked by the Wilcoxon rank sum test,
and the significance of difference in ratio by the chisquare test. Survival rates were calculated by the
Kaplan-Meier method. In this calculation, deaths attributable to diseases other than recurrence of lung cancer were excluded. Statistical evaluation in relation to
categories of prognostic factors was done with the log
rank test. The level of statistical significance was set at
P < 0.05.
Results

Distribution of DT of Lung Cancer

The minimum DT was 30 days, and the maximum DT
was 1077 days. DT after logarithmic conversion is
shown in Figure 2. Both skewness (0.7204)and kurtosis
(-0.0643) are small. Thus, DT was found to have a log
normal distribution (AM f SD of DT was 163.7 f 177.5
days, and GM was 113.3 days). The mean of DT of each
category of prognostic factors in terms of AM +- SD and
GM are shown in Table 1.

Comparison of Survival Rate in Relation to

Categories in Prognostic Factors (Univariate
Analyses)
DT, the distribution of which is log normal, was classified into two groups with reference to GM (113.3 days).
Tumors with a DT of less than 113.3 days were grouped

Tumor Doubling Time of Primary Lung CancerjUsuda et al.

2241

as rapidly growing tumors, and those with a DT of more

than 113.3 days as slowly growing tumors. The 5-year
survival rate (29%) of 98 patients with a rapidly growing tumors was significantly lower than that (55%) of
76 patients with slowly growing tumors (Fig. 3).
All of the patients of this study were classified into
two groups with reference to the mean age (67 years).
The 5-year survival rate (48%) of 82 patients who were
67 years or younger was significantly higher than that
(33%) of 92 patients who were older than 67 years. The
5-year survival rate (32%) of 123 men was significantly
lower than that @go/,)of 51 women. The 5-year survival rate (30%) of 45 patients with disease detected in a
hospital was significantly lower than that (44%) of 129
patients with disease detected by the screening trial.
The 5-year survival rate (35%) of 101 patients with a
smoking history was significantlylower than that (48%)
of 73 patients without a smoking history. The 5-year
survival rate (30%) of 85 patients with symptoms was

30L

1.5

2.0

2.5

3.0

(31.6)

(100)

(316)

(1 000)

LcglOScale
(days)

Log (DT)
Figure 2. Distribution after logarithmic conversion of DT. Skewness:
0.7204; kurtosis: -0.0643.

Table 1.Mean Tumor Doubling Time in Relation to Prognostic Factors

Mean DT (days)
Prognostic
factor
Sex
Detection
Smoking history
Symptoms
Therapy
Cell type

T factor

N factor

Men
Women
Detected by the screening trial
Detected when seen in hospital
With
Without
With
Without
Resection
Nonsurgical treatment
Adenocarcinoma
Squamous cell ca.
Small cell ca.
Large cell ca.
T1
T2
T3
T4

NO
N1
N2
N3

M factor

MO
M1

Stage

I
II
I11

Iv

GM

AMZSD

Category

(n = 123)
(n = 51)

(n = 129)
(n = 45)
(n = 101)
(n = 73)
(n = 85)
(n = 89)
(n = 126)
(n = 48)
(n = 86)
(n = 67)

(n = 7)
(n = 12)

(n = 63)
(n = 79)
(n = 19)
(n = 13)
(n = 86)
(n = 24)
(n = 54)
(n = 10)
(n = 158)
(n = 16)
(n = 73)
(n = 15)
(n = 70)
(n = 16)

126.4 t 133.5
253.5 2 232.1
162.5 ? 174.3
167.1 t 188.4
118.1 2 110.3
226.7 2 227.7
120.4 2 110.5
205.0 ? 216.2
175.3 2 190.3
133.0 ? 135.5

186.0
112.4
115.9

I*
IS

161.1
142.4
118.2
101.5

104.7 2 105.6
80.8 t 49.7
79.4 51.9

69.2
66.8

145.6 ? 177.8
139.0 Z 193.7

89.4

199.2 ? 207.5
135.0 -C 172.4
130.8 Z 127.2
104.1 55.5
165.2 ? 175.4
148.5 ? 202.4
212.7 -C 213.6 t
118.6 t 76.7
125.7 -t 129.4
148.5 2 202.4

132.8
93.7
99.4

93.2
115.0
98.3
144.3 t
100.3
93.4
98.3

AM 2 SD: arithmetic mean ? standard deviation; G M geometric mean; Ca: carcinoma; T factor: primary tumor
factor; N factor: regional lymph node factor; M factor: distant metastasisfactor.
* P < 0.001.

i P < 0.01.
$ P < 0.05.

CANCER October 25, 2994, Volume 74, No. 8

2242
1%)

iO0' b.

80

(n=76)

1
Rapidly-growing (DT< 11 3.3 days)
(n=98)

pco.001

_I

Years after diagnosis

Figure 3. Survival rate of patients with lung cancer as classified by

growth rate.

significantly lower than that (52%) of 89 patients without symptoms. The 5-year survival rate (2%) of 48 patients who underwent nonsurgical treatment was significantly lower than that (55%) of 126 patients who
underwent resection.
The 5-year survival rate was 47% for 86 patients
with adenocarcinoma, 37% for 67 patients with squamous cell carcinoma, 28% for 7 patients with small cell
carcinoma, and 16% for 12 patients with large cell carcinoma. The survival rate of patients with adenocarcinoma was significantly higher than that of patients with
squamous cell carcinoma and that of patients with large
cell carcinoma. The 5-year survival rate (62%) of 63 patients with a T1 lung cancer was significantly higher
than the 5-year survival rate (32%) of 79 patients with
a T2 lung cancer, than the 4-year survival rate (23%) of
19 patients with a T3 lung cancer and the 4-year survival rate (15%) of 13 patients with a T4 lung cancer.
The 5-year survival rate (74%) of 86 patients with an
NO lung cancer was significantlyhigher than the 3-year
survival rate (190/) of 24 patients with an N 1 lung cancer, the 3-year survival rate (12%) of 54 patients with
an N2 lung cancer, and the 2-year survival rate (0%) of
10 patients with an N3 lung cancer. The 2-year survival
rate (6%) of 16 patients with an M1 lung cancer was
significantly lower than the 5-year survival rate (45%)
of 158 patients with an MO lung cancer. The 5-year survival rate (80%)of 73 patients with a Stage I lung cancer
was significantly higher than the 2-year survival rate
(530/0)of 15 patients with a Stage I1 lung cancer, the 2year survival rate (28%) of 70 patients with a Stage 111
lung cancer, and the 2-year survival rate (6%) of 16 patients with a Stage IV lung cancer.

Multivariate Analyses Using the Cox Proportional

Hazard Model
Ten prognostic variables with reference to which survival rate was compared as described were classified as

follows: DT, continuous; age, continuous; sex (men = 0,

women = 1);detection (detected by the screening trial =
0, detected when seen in hospital = 1);smoking history
(without = 0, with = 1); symptoms (without = 0, with
= 1);therapy (resection = 0, nonsurgical treatment = 1);
primary tumor (T) factor (T1 = 0, T2 = 1, T3 = 2, T4 =
3); regional lymph node (N) factor (NO = 0, N1 = 1, N2
= 2, N3 = 3) and distant metastasis (M) factor (MO = 0,
M1 = 1).
The risk of death related to DT and other prognostic
factors was evaluated by multivariate analyses using
the Cox proportional hazard model. A final set of five
significant variables was obtained in a stepwise fashion
(Table 2): N factor ( P = 0.0001); therapy (P = 0.0016);
M factor (P = 0,0017); T factor (P = 0.0018); and DT (P
= 0.0152). The correlation coefficient was -0.193 between DT and N factor; -0.107 between DT and therapy; -0.027 between DT and M factor; and -0.134 between DT and T factor. As suggested by these small correlation coefficients, DT is an independent and
significant prognostic factor of lung cancer. However,
it was found that sex was not a significant factor (P =
0.0724).
The results of univariate analyses showed a significant dlfference in survival rates in relation to age,
sex, detection (how a tumor was detected), smoking history, and symptoms. But the results of multivariate
analyses showed that these five categories of prognostic
factor were not significant.

Discussion
Patients with lung cancer included in this study were
found in a limited number of 46 local municipalities
during a limited period to reduce selection bias and to
determine the exact distribution of DT as much as possible. DT was found to have a log normal distribution
after logarithmic conversion in accordance with the results of other investigator^.'^-'^ Thus, it was justified to
analyze the data of DT by using geometric means,"
rather than the arithmetic means that had been used for
the analyses of DT of lung cancer.
The minimum size of lung cancer that can be detected on a chest X-ray film has been reported to be 6
mm,1,3,11 7 mm,19,20 8 mm ,21 or 1 cm.7,22~illington'
stated that, if prior chest films show no tumor, one can
calculate an upper limit doubling time by assuming that
a nodule was present and was only 8 mm in greatest
dimension (the lower margin of detectability) in the
most recent negative film. In our experience, detection
of lung cancer that is smaller than 6 mm has been impossible. Thus, DT was calculated under the presumption that a tumor was present and had a greatest dimension of 6 mm in the most recent negative film taken
within 1 year before the time when the tumor was de-

Tumor Doubling Time of Primary Lung Cancer/Usuda et al.

2243

Table 2. Results of Multivariate Analyses Using the Cox Proportional Hazard Model
X2

Variable

Beta

Standard error

(Chi-sauare)

P-value

N factor

0.7485
0.8132
0.9816
0.3736
-0.0022
-0.4961

0.1212
0.2583
0.3125
0.1194
0.0009
0.2762

38.12

0.0001
0.0016
0.0017

Therapy
M factor
T factor

DT
Sex

9.91
9.87
9.79
5.89
3.23

0.0018
0.0152
0.0724

N factor: regional lymph node factor; M factor: distant metastasis factor; I factor: primary tumor factor

tected for the first time. Not all of the areas on chest Xray are as sensitive. However, all of the patients included in our study had a tumor shadow that was easier
to detect in the peripheral lung field. Thus, our assumption is valid that the minimum size of lung cancer that
can be detected on a chest X-ray film is 6 mm. If the
rapidly growing tumors are excluded from the analysis,
the distribution of DT may shift to the range of longer
DT, so the population included for evaluation may not
reflect the actual pattern of distribution of DT. Thus,this
assumption is justified in the absence of other suitable
methods.
We have reported previously that the prognosis of
the patients with a rapidly growing lung cancer is worse
than that of patients with a slowly growing lung can~ e r ' , ' and
~ that an early peripheral lung cancer24is undetectable in its initial phase but remains in the early
stage because the growth rate is
Gomperzian
kineticsz6predicts that the more advanced the tumor,
the more slowly it is expected to grow. However, our
result showed that patients with disease detected at advanced stages were expected to have shorter DT and
patients with disease detected at early stages were expected to have longer DT. In our study, the mean of
DT was significantly shorter in men than in women; in
patients with a smoking history than in patients without
a smoking history; and in patients with symptoms than
in patients without symptoms. The mean DT in patients
with an adenocarcinoma was significantly longer than
that in patients with a squamous cell carcinoma and significantly longer than that of patients with an undifferentiated carcinoma. The mean DT in patients with a T1
lung cancer was significantly longer than that in patients with a T2, T3, or T4 lung cancer. The survival rate
in patients with a shorter DT was significantly lower
than that in those with a longer DT. A similar tendency
was observed for other prognostic factors. The prognosis of patients with a shorter DT has been reported to be
poor~Z-4,7,15,1 6.23.27-30
In addition, the results of our study
showed that, even when analyzed with reference to
each prognostic factor, the survival rate of patients with
a shorter DT was significantly lower than that of those

with a longer DT. Thus, it was found that DT is closely

related to survival rate.
Univariate analyses showed significant differences
in survival in relation to DT, age, sex, detection (how a
tumor was detected), smoking history, symptoms, therapy, cell type, T factor, N factor, M factor, and stage,
but it is unclear which factors affect the survival rate
most strongly. In fact, there have been few publications
on the influence of DT on survival rate with reference
to TNM classification or prognostic factors other than
TNM classification. In multivariate analyses using the
Cox proportional hazard model for 10 factors (DT, age,
sex, detection, smoking history, symptom, therapy, T
factor, N factor, and M factor), five significant factors
affecting survival were selected in a stepwise fashion in
the increasing order as follows: N factor (P = O.OOOl),
therapy (P = 0.0016), M factor (P = 0.0017), T factor (P
= 0.0018), and DT (P = 0.0152). The correlation coefficient was -0,193 between DT and N factor, -0.107 between DT and therapy, -0.027 between DT and M factor, and -0.134 between DT and T factor. These four
correlation coefficients being small shows that DT is an
independent prognostic factor. TNM classification is
useful for the accurate estimation of prognosis. Thus, a
combination of DT and TNM classification is even more
useful for the same purpose.
The survival rate of patients with disease detected
in a hospital was significantly lower than that of patients with disease detected by the screening trial, and
the results of multivariate analyses showed that the factor of detection was not significant. These data may suggest length bias in cancer screening. The survival rate of
men was significantly lower than that of women, and
the results of multivariate analyses showed that the factor of sex was not significant. A correlation was found
between sex and other prognostic factors. Better SUP
vival in women was associated with long DT, not smoking, NO, resection, and absence of symptoms. Factors of
age, smoking history, and symptoms were significantly
correlated with survival when they were analyzed by
univariate analyses, but multivariate analyses showed
that they were not significantly correlated.

2244

CANCER October 15,1994, Volume 74,No. 8

Growth rate data help to differentiate between benign and malignant pulmonary nodules.31 Spratt and
Spratt31 reported that a tumor with a DT slower than
500 days usually was benign. In our study, 162 (93%)
of 174 patients had DT of 30-500 days, in contrast with
the data by Spratt and Spratt. In our study, GM (AM) of
DT in cell type was 163.3 days (223.1 days) for adenocarcinomas, 80.3 days (104.8 days) for squamous cell
carcinoma, 69.2 days (80.9 days) for small cell carcinomas, and 66.8 days (79.4 days) for large cell carcinomas.
Growth rate has been reported to be faster in squamous
cell carcinomas and undifferentiated carcinomas but
slower in a d e n o c a r c i n ~ m a s . ~Filderman
, ~ , ~ ~ ~ ~et
~ aL7 reported that the AM of DT was 180 days for adenocarcinomas, 100 days for squamous cell carcinomas and
large cell carcinomas, and 30 days for small cell carcinomas. Geddes4 compiled data from 228 patients for
whom DT were available in the literature and reported
that the AM of DT was 161 days for patients with adenocarcinomas, 88 days for those with squamous cell
carcinomas, 86 days for those with large cell carcinomas, and 29 days for those with small cell carcinomas.
According to Fujimura et
the AM of DT was 116
days for adenocarcinomas, 94 days for squamous cell
carcinomas, and 71 days for large and small cell carcinomas. This difference in mean DT may be attributable
to a difference in the protocols by which subjects for
evaluation were selected.
References
1. Spratt JS, Ter-Pogossian M, Long RTL. The detection and
growth of intrathoracic neoplasms. Archiv Surg 1963; 86:283-8.
2. Meyer JA. Growth rate versus prognosis in resected primary
bronchogenic carcinomas. Cancer 1973; 31:1468-72.
3. Weiss W. Tumor doubling time and survival of men with bronchogenic carcinoma. Chest 1974; 65:3-8.
4. Geddes DM. The natural history of lung cancer: a review based
on rates of tumour growth. BrJDis Chest 1979; 73:l-17.
5. Mizuno T, Masaoka A, Ichimura H, Shibata K, Tanaka H, Niwa
H. Comparison of actual survivorship after treatment with survivorship predicted by actual tumor-volume doubling time from
tumor diameter at first observation. Cancer 1984; 532716-20.
6. Heikkila L, Mattila P, Harjula A, Suomalainen RJ, Mattila S. Tumour growth rate and its relationship to prognosis in bronchioloalveolar and pulmonary adenocarcinoma. Ann Chir Gynaecof
1985; 74:210-4.
7. Filderman AE, Shaw C, Matthay RA. Lung cancer: Part I. Etiology, pathology, natural history, manifestations, and diagnostic
techniques. Invest Radiol 1986; 21230-90.
8. Usuda K, Saito Y, Takahashi S, Kanma K, Sagawa M, Sato M, et
al. Clinical study using previous miniature radiographs of lung
cancer cases detected during radiologic screening program. 1Jpn
Lung Cancer SOC 1990; 30:857-61.
9. Lillington GA. Management of solitary pulmonary nodules. In:
Bone RC, editor. Disease-a-Month. Littleton: Mosby-Year Book,
199 1970-3 18.
10. Schwartz M. A biomathematical approach to clinical tumor
growth. Cancer 1961; 14:1272-94.

11. Arai T, Shiobara J, Shiozawa M, Iwai K. Discussion of the problems of early detection of peripheral lung cancer from the standpoint of the growth rate and the size of the tumor at detection. ]
Jpn Lung Cancer SOC 1976; 16:7-13.
12. Cox DR. Regression models and life-tables. J R Stat Soc (B) 1972;
34:187-220.
13. The Japan Lung Cancer Society. General rules for clinical and
pathological record of lung cancer, ed. 3. Tokyo: Kanehara,
1987.
14. International Union Against Cancer. TNM classification of malignant tumors, ed. 4. Berlin: Springer-Verlag, 1987.
15. Spratt JS, Spratt TL. Rates of growth of pulmonary metastases
and host survival.Ann Surg 1964; 159:161-71.
16. Chahinian AP, Israel L. Rates and patterns of growth of lung
cancer. In: Israel L, Chahinian AP, editors. Lung cancer: natural
history, prognosis, and therapy. New York: Academic Press,
1976~63-79.
17. Charbit A, Malaise EP, Tubiana M. Relation between the pathological nature and the growth rate of human tumors. Eurf Cancer
1971; 7~307-15.
18. Tango T, Furukawa T. Medical statistics. Tokyo: Asakura Shoten, 1983.
19. Matsuda M, Horai T. A roentgen study of the small peripheral
lung cancer. J l p n Lung Cancer Soc 1971; 11:35-45.
20. Heelan RT, Flehinger BJ, Melamed MR, Zaman MB, Perchick
WB, Caravelli JF, et al. Non-small-cell lung cancer: results of the
New York screening program. Radiology 1984; 151289-93.
21. Steele JD, BueIl P. Asymptomatic solitary pulmonary nodules:
host survival, tumor size, and growth rate. 1Thor Cardiovasc Surg
1973; 65:140-51.
22. Straus MJ, Janis MG, Moran RE. Tumor biology of lung cancer.
In: Harris CC. Pathogenesis and therapy of lung cancer. New
York Marcel Dekker, 1978:611-51.
23. Usuda K, Saito Y, Sagawa M, Sat0 M, Nagamoto N, Fujimura S,
et al. Clinical and radiographic changes in relation to times of
screening and survival in lung cancer cases detected by the Miyagi screening program using annual chest radiographs. J Jpn
Lung Cancer SOC 1990; 30:483-90.
24. Ikegda S. Atlas of early cancer of major bronchi. Tokyo: Igaku
Shoin, 1976.
25. Usuda K, Takahashi 5, Kanma K, Ota 5, Imai T, Saito Y, et al.
Analysis of the diagnostic process and a retrospective study of
the Miyagi annual chest X-ray lung cancer screening program. J
Jpn Lung Cancer Soc 1988; 28:343-52.
26. Gompertz B. On the nature of the function expressive of the law
of human mortality, and on a new mode of determining the
value of life contingencies. Philos Trans R SOC 1825; 115:513-85.
27. Malaise EP, Chavaudra N, Charbit A, Tubiana M. Relationship
between the growth rate of human metastases, survival and
pathological type. EurJ Cancer 1974; 10:451-9.
28. Bone RC, Balk R. Staging of bronchogenic carcinoma. Chest
1982; 821473-80.
29. Joseph WL, Morton DL, Adkins PC. Prognostic significance of
tumor doubling time in evaluating operability in pulmonary
metastatic disease. 1Thorac Cardiovasc Surg 1971; 6123-32.
30. Collins VP, Loeffler RK, Tivey H. Observations on growth rates
of human tumors. Am J Roentgenof 1956; 76:988-1000.
31. Spratt JS, Spratt JA. The prognostic value of measuring the gross
linear radial growth of pulmonary metastases and primary pulmonary cancers. 1 Thorac Cardiovasc Surg 1976; 71274-8.
32. Weiss W. Peripheral measurable bronchogenic carcinoma. Am
Rev Respir Dis 1971; 103:198-208.
33. Fujimura S, Suda S, Yamauchi A, Sato H, Sohara Y, Kondo T, et
al. Tumor doubling time and PPD skin test reactivity in resectable lung cancer. J J p n Lung Cancer Soc 1979; 19:135-42.