Contemporary Clinical Trials 29 (2008) 720726

Contents lists available at ScienceDirect

Contemporary Clinical Trials

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c o n c l i n t r i a l

Analyzing biological rhythms in clinical trials

Naser B. Elkum a,,1, James D. Myles b,1, Pranesh Kumar c,1
a
King Faisal Specialist Hospital and Research Center, Kingdom of Saudi Arabia
b
Michigan Institute for Clinical and Health Research, University of Michigan, USA
c
University of Northern British Columbia, Prince George, BC, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Background: The human body exhibits a variety of biological rhythms. There are patterns that
Received 20 November 2007
correspond, among others, to the daily wake / sleep cycle, a yearly seasonal cycle and, in
Accepted 13 May 2008
women, the menstrual cycle. Sine/cosine functions are often used to model biological patterns
for continuous data, but this model is not appropriate for analysis of biological rhythms in
Keywords:
Biological rhythms
failure time data.
Failure time data Methods: We consider a method appropriate for analysis of biological rhythms in clinical trials.
Cosinor rhythmometry We present a method to provide an estimate and condence interval of the time when the
Weibull distribution minimum hazard is achieved. A motivating example from a clinical trial of adjuvant of pre-
Cox model menopausal breast cancer patients provides an important illustration of the methodology in
Menstrual cycle
practice.
Breast cancer
Results: Adapting the Cosinor method to the Weibull proportional hazards model is proposed as
useful way of modeling the biological rhythm data. It presents a method to estimate the time
that achieves the minimum hazard along with its associated condence interval. The
application of this technique to the breast cancer data revealed that the optimal day for pre-
resection incisional or excisional biopsy of 28-day cycle (i.e. the day associated with the lowest
recurrence rate) is day 8 with 95% CI 510. We found that older age, fewer positive nodes,
smaller tumor size, and experimental treatment are important prognostic factors of longer
relapse-free survival.
Conclusions: The analysis of biological/circadian rhythms is usually handled by Cosinor
rhythmometry method. However, in FTD this is simply not possible. In this case, we propose to
adapt the Cosinor method to the Weibull proportional hazard model. The advantage of the
proposed method is its ability to model survival data. This method is not limited to breast
cancer data, and may be applied to any biological rhythms linked to right censored data.
2008 Elsevier Inc. All rights reserved.

1. Introduction rhythmometry, in which a linear least squares regression is

Various mathematical models have been used to assess
the suitability of periodic functions associated with biological
rhythms. The most common approach is that of Cosinor
Corresponding author. King Faisal Specialist Hospital and Research
Centre, MBC 03-BESC, P.O. Box 3354, Riyadh, 11211, Kingdom of Saudi Arabia.
Tel.: +966 1 464 7272x32561; fax: +966 1 442 4542.
E-mail address: nkum@kfshrc.edu.sa (N.B. Elkum).
1
The authors contributed equally to this work.
used to t a sinusoidal curve to time-series data [14]. Others
have considered a nonparametric smooth curve based on
tting a periodic spline function for human circadian rhythms
[5,6]. These approaches, however, are intended to study
measurements over time from multiple subjects to study the
inherent dynamics of circadian rhythms. Although these
curve-tting techniques can be helpful, they are not suitable
for right censored failure time data (FTD).
Failure may be broadly dened as the occurrence of a
pre-specied event. Events of this nature include time of

1551-7144/$ see front matter 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.cct.2008.05.001
 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726
death, disease occurrence or recurrence and remission. One
important aspect of FTD is that the anticipated event may
not occur for each individual under study. This situation is
referred to as censoring and the study subject for which no
failure time is available is referred to as censored. Censored
data analysis requires special methods to compensate for
the information lost by not knowing the time of failure
of all individuals. There is a scarcity of literature about
the methodologies to deal with circadian or biological
rhythms in failure time data. Elkum and Myles [7] proposed
modeling biological rhythms in failure time data by adapt-
ing the Cox model. However, the Cox proportional hazard
(CPH) model, does not assume any particular hazard func-
tion or distribution of survival time data. If the assump-
tion of particular probability distribution for the data is
valid, estimates of quantities like relative hazards, median
survival times will have smaller standard errors than they
would have under the assumption of nonparametric hazard
function.
This article models the biological rhythms in censored
data. It presents a method to estimate the time that achieves
the minimum hazard along with its associated condence
interval. The model is then used to predict the optimal time in
the menstrual cycle for breast cancer surgery (i.e. day asso-
ciated with the lowest recurrence rate) in pre-menopausal
women using data from the National Cancer Institute of
Canada Clinical Trial Group MA.5 study.
2. Methods
For modeling biological rhythms in failure time data,
we consider a method that consists of modeling the ha-
zard function using the Weibull distribution and then using
the Weibull hazard function in the Cosinor rhythmometry
model.
2.1. The model
The general approach to analyze biological rhythm is that
of Cosinor Rhythmometry (Nelson et al. [3]). The Cosinor
model for analyzing data span is the best-tting cosine func-
tion of the form:
f si M Acossi / ei 1 The CosinorWeibull proportional hazard (CWPH) model
where si represents the time of measurements or interven-
tions for ith individual, M the mean level (termed mesor) of
the cosine curve, A is the amplitude of the function, is the
angular frequency (period) of the curve, and is the acro-
phase (horizontal shift) of the curve. It is assumed that the
errors, i, are independent and normally distributed with
means zero and a common residual variance 2. It is also
possible to use more than one cosine function with different
values of (whether or not in harmonic relation) or a
combined linearnonlinear rhythmometry [8,9]. The con-
ventional method of least squares analysis is applied to
estimate the constants of the model (1). However, a problem
with implementing this method is that the assumptions
about errors may be invalid. For failure time data, in parti-
cular, with skewed and censored observations the error
assumptions do not hold. The important inference question
721
in this setting is about the conditional distribution of failure,
given the covariates.
The Weibull proportional hazard (WPH) model is an
appropriate method for regression analysis of survival data.
The WPH model imposes a parametric distributional form on
the hazard function by modeling the survival times as a
Weibull distribution. The WPH model is described mathe-
matically as follows.
Let XT = (X1,, Xp) denote the p-measured covariates on
individuals enrolled for the study with censored failure time
observations. Then the WPH model for the study individuals
is written as
 


t t 1 exp T X ; 2
where T = (X1,, Xp) is a vector of regression coefcients, and
TX is an inner product. The survival time has a Weibull
distribution with scale parameter exp(TX) and shape
parameter . The effect of the covariates in the model is to
alter the scale parameter while the shape parameter remains
unchanged. In order to examine the effect of biological rhythm
upon survival using this model, one needs to adapt the Cosinor
rhythmometry to the WPH model.
Let us assume that we have n independent individuals
(i = 1, n). For each individual i, the survival data consist of
the time of the event or the time of censoring ti, an indicator
variable i with a value of 1 if ti is uncensored or a value of 0 if
ti is right censored such that ni1 i r, where r is the number
of deaths, Xi = (X1i,, Xpi), and Yi = (Y1i,Y2i), so that the ob-
served data are (ti,i,Xi,Yi). We dene Y1i = cossi, Y2i = sin si,
= 2/, with being the cycle span, and 1 = Acos and 2 =
Asin. A cosine varies between +1 and 1, so its coefcient
will represent half of the amplitude of the cycle. In the case of
a survival function this represents half of the difference in
relapse between patients who underwent surgery on days
of maximum and minimum risk. The periodicity is easily
tested using the sinusoidal curve, where the signicance of
Y1 or Y2 indicates periodicity. Note that, since Sin(2s/) =
Y2 this means that the time of surgery s = (/2)sin 1(Y2).
The angular frequency () must be set based on prior know-
ledge of the pattern. For Circadian rhythms this is usually
24 hours and for longer biological rhythms this can differ from
individual to individual.
for the ith of the n individuals, i = 1,, n, can be written as
h
  i
i t ti1 exp T Xi exp1 Y1i 2 Y2i 3
It should be noted that the effects in this model are
multiplicative instead of additive as in the linear model case.
When = 1, we get a special case of the CWPH model (3).
We call it the Cosinor-exponential proportional hazard (CEPH)
model which may be written as
h  i
i t exp T Xi exp1 Y1i 2 Y2i : 4
This can be used instead of CWPH in situation where data
are best t with an exponential model. However CWPH is a
more general model. It will reduce to the exponential model
when the data is exponential.
722 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726

2.2. Estimation of the CWPH model parameters The asymptotic 95% condence interval of optimal time
will be based
 on the standard errors
r using an assumption of
 
The CWPH model in Eq. (3) is tted by constructing the
normality Smin FZ =2 var Smin .

likelihood function of the n observations using ni1 ff ti g i
fSti g1i , and maximizing this function with respect to the To characterize the optimal time for intervention, we t a
unknown parameters, T, , , 1 and 2. This can be done parametric model to the estimated optimal times empirically.
using computer software for survival analysis. The software
output generates the standard errors of the parameter 3.1. Motivating application: biological timing of breast
estimates from which condence intervals can be calculated. cancer surgery
Let , ,
T, 1 and 2 be the estimates of the parameters in the
model (3). Then, the estimate of is obtained by reconverting While seasonality affects us all, the menstrual cycle


the estimated 1 and 2 as / tan1 2 . The variance of is directly affects about 52% of the world's inhabitants. Each of
1
calculated using the delta method [10] and is shown to be: the members of this small global majority spends about half
of her life participating regularly and continuously in this
  2
 2

 powerful biological rhythm. Many diverse disease activities
var
1
1 var 2 2 2 cov
1 1 ;
2
var / 2 h 2 i 2
: 5 have been demonstrated to be affected by this cycle. Cancer is
2
1
2 one of these [1114]. It has been conjectured that menstrual
stage at time of resection might affect breast cancer outcome.
An approximate
r (1 )100% condence interval of is Recurrence of breast cancer disease may be affected by timing
 
=
/FZ var / , where Z/2 is the (1 /2)100% percentile the surgery in relation to the menstrual cycle; therefore, the
2
timing of surgery may be an important element that affects
of the standard normal distribution.
breast cancer outcome [15,16].
The estimation
q
of the amplitude can be obtained by
Hrushesky et al. proposed that pre-menopausal patients
2 2
A 1
2 . The asymptotic variance of can also be ob- with breast cancer who underwent surgery during the luteal
tained using the delta method and is shown to be: phase of the menstrual cycle had higher disease-free and
overall survival rates than did patients operated on during
  2
 2


var 1
1 var 2 2 1 2 cov 1 ;
2 other phases of the cycle [17]. One biologicalendocrinologi-
var A 2 : 6
2
1
2
2 cal basis of this is that in the follicular phase (rst 14 days of
the cycle) estrogen reduces immune activity, phagocytic
An approximate
r (1 ) 100% condence interval of A is
  activity and circulating levels of IL-2 therefore potentially
=
AFZ var A .
2 increasing metastatic potential of breast cancer cells [18]. In
the luteal phase (second 14 days of the cycle) the increase in
3. Optimal time and its condence interval circulating levels of endogenous progesterone may modulate
the proliferation of normal and neoplastic breast tissue and
One may obtain the optimal time by trying different increase intercellular cohesion therapy reducing metastatic
partitions to the data where the variation looks cyclical. potential [19,20].
Evidently, the connection between survival and menstrual Using biological information, the timing of surgical
day of surgery is a nonlinear relationship. Therefore, in order intervention for breast cancer may have an inuence on the
to analyze the cyclic trend, we relied on the knowledge of outcome of such interventions [21,22]. These studies have
sinusoidal curve. The simplest cyclical pattern is the sine wave shown that patients who have surgery during the follicular
with its associated parameters of amplitude, mean level phase (rst 14 days of the cycle) have a higher recurrence rate
(mesor) angle frequency, and phase angle (acrophase). We than those treated during the luteal phase (second 14 days of
will construct an estimate and condence interval of the time the cycle). Other studies have shown that patients having
where the minimum hazard is achieved. The objective is to surgery during the perimenstrual period (days 06 and 21
locate the optimal time for intervention, which is the time 36) of the menstrual cycle had a quadrupled risk of recurrence
where the curve is at a minimum. and death compared with women operated upon during the
Since we know that cos = 1, therefore, the optimum is middle (days 7 to 20) of their menstrual cycle [17,23]. Badwe
when = t+ where = 2/. Hence, et al. [24] stratied patients into groups containing patients
" # whose LMP was 312 days before surgery and those who


were operated on at other times. They showed that overall
smin 0:5 7
2 and recurrence-free survivals were each enhanced for those
who were resected during the luteal phase. Moreover, even
The variance of Smin is the denitions of follicular and luteal phases were made
based on different criteria in different institutions. Any time
  2   an article was published there were subsequent letters to the
var Smin 2 var / editor or articles presenting contradictory results [2527].
4
These discrepancies might be explained by the limited
h 2


i
2 8 reliability of the menstrual history data, and the fact that
2
2 1
2 var 2 2
1 var 1 1 ;
2 cov
these studies were retrospective. This has stimulated our
h i interest in developing a more rigorous method to estimate the
2 2 2
42 2
1
(8) best time that can be recommended for surgical intervention
 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726 723

Table 1 Table 2
CWPH analysis of variables affecting disease-free survival Demographic and pathologic characteristics of 262 pre-menopausal patients

Variable Parameter p-value RR 95% CI Predictors Time of LMP (number)

estimate 510 Others
Age 0.0535 0.0187 0.9479 0.90660.9911 (n = 48) (n = 214)
Number of nodes 0.0994 0.0002 1.1045 1.04761.1646 Age [mean (SD)] 42.9 (5.6) 42.8 (5.5)
Treatment 1.0197 0.0002 0.3607 0.21066170 Nodes [median (SD)] 3.0(2.3) 3.0 (3.8)
Stage I 0.9582 0.0088 0.3836 0.18740.7853
Stage II 0.8507 0.0087 0.4271 0.22620.8063 Treatment
Stage III 1 CMF n (%) 22 (8.4) 112 (42.8)
Y2 = sin(2s/) 0.6881 b 0.0001 0.5025 0.36170.6982 CEF n (%) 26 (9.9) 102 (38.9)

based on prospective study. The best way to answer this
question would be to perform a randomized controlled
clinical trial as proposed by Badwe et al. [28] in their meta-
analysis of timing of surgery data.
The MA.5 study was a multi-center clinical trial con-
ducted by the National Cancer Institute of Canada Clinical
Trial Group (NCIC CTG) [29]. There were 262 pre-menopausal
patients who had adequate data for last menstrual period
(LMP) included in the study. All of them were eligible for the
study, had normal menstruation, and regular period cycles
(lasting between 21 and 35 days) [30]. The recurrence of
breast cancer was conrmed with clinical or pathologic
assessment, or both. Initial tumor size, status of the axillary
lymph nodes, and other prognostic factors were assessed
clinically and pathologically. Disease-free months were
calculated from date of surgery to date of rst relapse. The
trial was activated December 1, 1989 and closed to accrual on
July 31, 1993. The objective was to examine disease-free
survival in relation to the timing of breast tumor excision
during the menstrual cycle.
All analyses were conducted with SAS Version 9.1 and S-
Plus for Windows Version 6.0. All tests were two-sided, and a
Stage
I 18 (6.7) 88 (33.6)
II 25 (9.5) 103 (39.3)
III 5 (1.9) 23 (8.8)
level of = 0.05 was used to determine a signicant result.
Product-limit survival curves were calculated by the method
of KaplanMeier.
4. Results
Prior to tting the model assuming the Weibull form for
the survivor times, this assumption needs to be validated. For
validity, we t the function ln{lnS(t)} = ln + lnt, where S(t)
is the KaplanMeier estimate of the survivor function. The
result of the t was: ln{lnS(t)} = 3.1088 + 1.0474 lnt; R2 =
0.9569; p b 0.0001. Since the p-value is very small and the plot
is close to the straight line, assumption of the Weibull distri-
bution to model the survivor times is tenable.
The CWPH model identied age, positive nodes, patholo-
gic stage, and experimental treatment as the most signicant
factors related to the disease-free survival. The assumption of
proportional hazard was satised by CoxSnell residuals. The
estimates of the CWPH model parameters are given in Table 1.
A positive parameter estimate indicates an increasing hazard

Fig. 1. Relapse-free survival by timing of surgery: proposed denition.

724 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726

Table 3 Table 4
Optimal time for intervention for regular menstrual cycle KM summary of different times of breast cancer surgery criteria

Cycle Number Optimal 95% condence interval Criteria LMP days Relapse (%) p-value
length of women time of
Lower Upper CWPH 04 and 1140 59 (27.6) 0.0286
surgery
510 6 (12.5)
21 9 5.67 3.84 7.51 CPH 03 and 1340 55 (30) 0.0084
22 5.95 4.02 7.87 412 10 (13)
23 3 6.22 4.21 8.23 Follicular 014 28 (20) 0.070
24 6 6.49 4.39 8.58 Luteal 1540 37 (30)
25 13 6.76 4.57 8.94 Mid-cycle 720 23 (20) 0.062
26 10 7.03 4.75 9.30 Perimenstrual 06 and 2140 42 (29)
27 8 7.30 4.94 9.66
28 167 7.57 5.12 10.01
29 2 7.84 5.3 10.37
30 31 8.11 5.49 10.73 as 1.2286. This estimated Weibull model is plotted in Fig. 2.
31 3 8.38 5.67 11.09 The gure exhibits the graphical presentation of the estimated
32 4 8.65 5.85 11.44 model for times of intervention. This additionally can be used to
33 8.92 6.03 11.80
calculate summary characteristics of times of intervention or
34 1 9.19 6.22 12.16
35 1 9.46 6.40 12.52 predict probabilities given the times of intervention.
Since different comparisons had been made in the past
based on retrospective analyses, it was of interest to compare
results from approaches used in the past with the approach
with increasing values of the covariate. A negative value proposed in this study. The information was summarized in
indicates that the hazard decreases with increasing values of Table 4. The table evaluates disease-free survival for the
the covariates. The hazard increases with increasing number of previous used biological criteria and our proposed criterion
positive nodes, as seen by the positive parameter estimate of (CWPH) for optimal time of breast cancer surgery.
0.0994.
Assuming the length of the menstrual cycle is 28 days for 4.1. Follicular vs. luteal stage comparison
all women, and using back-transformation, we obtained the


acrophase / tan1 0:6985
0:0898 1:443. Therefore, the minimum
The risk for recurrence differed between the two phases:

 30% of patients developed recurrence after surgery in the
time of the curve is Smin 14 1 1:44
3:14 7:578 days. The
luteal group compared with 20% in the follicular group. Fig. 3
variance was estimated using Eq. (8) as 1.559 and hence the
indicates a higher recurrence rate for patients with tumor
95% condence lies between 5 and 10 days.
excision during the luteal phase, but the differences in sur-
Two hundred and fourteen pre-menopause women (81.9%)
vival were not statistically signicant (p = 0.07).
had LMP during 04 and 1140 days, and 48 women (18.1%)
had LMP in 510 days. Using this optimal interval, the disease
4.2. Mid-cycle vs. perimenstrual stage comparison
recurred in 59 patients (27.6%) in the group where LMP was
04 and 1140 days, 6 patients (12.5%) developed disease
In this kind of menstrual interval, tumor recurred in 29% of
in mid-cycle (510 days) group. Fig. 1 shows a statistical sig-
perimenstrual patients and in 20% of mid-cycle patients. Fig. 4
nicant difference in survival between these two groups
indicates statistically insignicant difference in time to rst
(p = 0.0286). Based on regular menstrual cycle of lengths 21
recurred by phase (p = 0.062).
36 days, Table 2 shows the distribution of the predictors across
the proposed optimal timing of surgery. Table 3 depicts optimal
5. Conclusion
timing of surgery for different menstrual cycles. The optimal
times for intervention followed the Weibull model with scale
This paper has presented a method for modeling failure
parameter estimated as 8.3515 and shape parameter estimated
time data with an underlying biological rhythm. The advan-
tage of adapting a Cosinor model to the Weibull proportional
hazard model is its ability to model right censored data. We
have presented a method to provide an estimate and con-
dence interval of the time where the minimum hazard is
achieved.
The application of this technique to breast cancer data
revealed that the optimal days for pre-resection incisional or
excisional biopsy of 28 day cycle (i.e. the days associated with
the lowest recurrence rate) are days 510. This represents the
putative follicular phase for women with 28 to 36 day cycle
duration and the luteal phase for those with the usual cycle
length between 21 and 28 days. This is in agreement with the
contention that disease recurrence and metastasis are more
frequent and appear more rapidly in women who have had
their initial breast cancer resection during days 06 and 21
Fig. 2. Estimated Weibull model for the optimal times of intervention. 36 of the menstrual cycle.
 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726 725

Fig. 3. Relapse-free survival by timing of surgery: Senie's denition.

The question of the best time for surgery has been debated
in the cancer literature and remains an open question. The
results from the MA.5 trial contradict some of the other
data published about this. A full debate of this question is
outside the scope of this paper. The proposed analytical
technique is not limited to breast cancer data and may be
applied to any biological rhythms linked to right censored
data.
Deciding upon a specic model depends on data used,
sample size as well as selecting a shape that best depicts the
true hazard function. If the assumption of a particular pro-
bability distribution for the data is valid, estimates of quan-
tities like relative hazards, median survival times will have
smaller standard errors than they would have under the
assumption of nonparametric hazard function. Most statistical
software (using probability plots) allows to assess whether the
chosen functional form is correct for a given data set.
Competing interests
We did not identify any situation that might be perceived
as a conict of interest.

Fig. 4. Relapse-free survival by timing of surgery: Hrushesky's denition.

726 N.B. Elkum et al. / Contemporary Clinical Trials 29 (2008) 720726
References
[1] Tong YL. Parameter estimation in studying circadian rhythms. Bio-
metrics 1976;32(1):8594.
[2] Ware JH, Bowden RE. Circadian rhythm analysis when output is
collected at intervals. Biometrics 1977;33(3):56671.
[3] Nelson W, Tong YL, Lee JK, Halberg F. Methods for Cosinor-rhythmo-
metry. Chronobiologia 1979;6(4):30523.
[4] Hickey DS, Kirkland JL, Lucas SB, Lye M. Analysis of circadian rhythms by
tting a least squares sine curve. Comput Biol Med 1984;14(2):21723.
[5] Wang Y, Brown MB. A exible model for human circadian rhythms.
Biometrics 1996;52(2):58896.
[6] Wang Y, Ke C, Brown MB. Shape-invariant modeling of circadian
rhythms with random effects and smoothing spline ANOVA decom-
positions. Biometrics 2003;59(4):80412.
[7] Elkum NB, Myles JD. Modeling biological rhythms in failure time data.
J Circad Rhythms 2006;4:14.
[8] Halberg F. Chronobiology: methodological problems. Acta med rom
1980;18:399440.
[9] Cornelissen G, Halberg F. Chronomedicine. Chicheter, UK: John Wiley &
Sons Ltd; 2005.
[10] Kendall M, Stuart A. The advanced Theory of Statistics, vol. I. London:
Grifn; 1986.
[11] Lis CG, Grutsch JF, et al. Circadian timing in cancer treatment: the
biological foundation for an integrative approach. Integr Cancer Ther
2003;2(2):10511.
[12] Levi F. Circadian chronotherapy for human cancers. Lancet Oncol
2001;2(5):30715.
[13] Hrushesky WJ. Rhythmic menstrual cycle modulation of breast cancer
biology. J Surg Oncol 2000;74(3):23841.
[14] Wood PA, Hrushesky WJ. Circadian rhythms and cancer chemotherapy.
Crit Rev Eukaryot Gene Expr 1996;6(4):299343.
[15] Demicheli R, Bonadonna G, Hrushesky WJ, Retsky MW, Valagussa P.
Menopausal status dependence of the timing of breast cancer
recurrence after surgical removal of the primary tumour. Breast Cancer
Res 2004;6(6):R689696.
[16] Zurrida S, Galimberti V, Gibelli B, et al. Timing of breast cancer surgery in
relation to the menstrual cycle: an update of developments. Crit Rev
Oncol Hematol 2001;38(3):22330.
[17] Hrushesky WJ, Bluming AZ, Gruber SA, Sothern RB. Menstrual inuence
on surgical cure of breast cancer. Lancet 1989;2(8669):94952.
[18] Hrushesky WJ, Gruber SA, Sothern RB, et al. Natural killer cell activity:
age, estrous- and circadian-stage dependence and inverse correlation
with metastatic potential. J Natl Cancer Inst 1988;80(15):12327.
[19] Navarrete MA, Maier CM, Falzoni R, et al. Assessment of the proliferative,
apoptotic and cellular renovation indices of the human mammary
epithelium during the follicular and luteal phases of the menstrual
cycle. Breast Cancer Res 2005;7(3):R30613.
[20] Soderqvist G, Isaksson E, et al. Proliferation of breast epithelial cells in
healthy women during the menstrual cycle. Am J Obstet Gynecol
1997;176(I Pt I):1238.
[21] Saad Z, Bramwell V, Duff J, et al. Timing of surgery in relation to the
menstrual cycle in premenopausal women with operable breast cancer.
Br J Surg 1994;81(2):21720.
[22] Senie RT, Rosen PP, Rhodes P, Lesser ML. Timing of breast cancer excision
during the menstrual cycle inuences duration of disease-free survival.
Ann Intern Med 1991;115(5):33742.
[23] Spratt JS, Zirnheld J, Yancey JM. Breast Cancer Detection Demonstration
Project data can determine whether the prognosis of breast cancer is
affected by the time of surgery during the menstrual cycle. J Surg Oncol
1993;53(1):49.
[24] Badwe RA, Gregory WM, Chaudary MA, et al. Timing of surgery during
menstrual cycle and survival of premenopausal women with operable
breast cancer. Lancet 1991;337(8752):12614.
[25] Goldhirsch A, Gelber R, Forbes J. Timing breast cancer surgery. Lancet
1991;338(8768):6913.
[26] Nomura Y, Kataoka A, Tsutsui S, Murakami S, Takenaka Y. Lack of
correlation between timing of surgery in relation to the menstrual cycle
and prognosis of premenopausal patients with early breast cancer. Eur J
Cancer 1999;35(9):132630.
[27] Milella M, Nistico C, Ferraresi V, et al. Breast cancer and timing of
surgery during menstrual cycle: a 5-year analysis of 248 premenopausal
women. Breast Cancer Res Treat 1999;55(3):25966.
[28] Badwe RA, Mittra I, Havaldar R. Timing of surgery during the menstrual
cycle and prognosis of breast cancer. J Biosci 2000;25(1):11320.
[29] Levine M, Bramwell V, Pritchard K, et al. Randomized trial of intensive
cyclophosphamide, epirubicin, and uorouracil chemotherapy com-
pared with cyclophosphamide, methotrexate, and uorouracil in pre-
menopausal women with node-positive breast cancer. J Clin Oncol
1998;16(8):26518.
[30] Fehring RJ, Schneider M, Raviele K. Variability in the phases of the
menstrual cycle. J Obstet Gynecol Neonatal Nurs 2006;35(3):37684.