Cancer Mortality Affected by the Choice of Anesthetic Drugs?
Purpose and aims By means of record-linkage, information from two registers will be merged
to retrospectively examine possible associations between survival from breast-, colorectal-,
or skin cancer with the choice of hypnotic used during surgical removal of the cancer.
We hypothesize that: the one- and five-year survival rate after radical breast-, colorectal-
or skin cancer surgery in general anesthesia is significantly higher in patients given the
intravenously administered hypnotic propofol compared with the survival in patients exposed
to the inhalational hypnotic sevoflurane.
The hypothesis is based on: 1) The knowledge about the opposite effects on the immune system
from the two different anesthetics and from their different genotoxic potentials 2) The
well-established associations between the state of the immune system and cancer growth, and
DNA damage and cancer development, with potential influences on survival.
Survey of the field Rationale and current state of knowledge A) Immuno-modulation Converging
evidence from animal studies and studies of human cell-lines indicate that different
anesthetics have opposite effects on the immune system. Commonly used inhalational hypnotics
are in this context pro-inflammatory, whereas the intravenously administered hypnotic agent
propofol is anti-inflammatory and also anti-oxidative. A few clinical studies have indicated
similar effects in patients, and a recent review has suggested that "tailoring an anesthetic
plan to patient's needs will become increasingly critical, and immunology should help in
More specifically, previous studies have investigated the immunological effects of different
anesthetics on monocytes, macrophages, natural killer cells, t-cytotoxic cells, and t-helper
cells. By affecting t-helper cells anesthetics indirectly affects the production of
anti-inflammatory mediators, such as interleukin-4 and -10. Anesthetics also affect the
production of pro-inflammatory cytokines, such as tumor necrosis factor alpha, and
interleukin-1 and -6. Moreover, the effects could be indirect by blocking or non-blocking of
the surgical stress response via the hypothalamic-pituitary-adrenal axis and the sympathetic
nervous system. Thus, stress hormones, such as catecholamines and cortisol, mediates
inhibitory effects on immune functions. In a highly complex way, the neuroendocrine system
together with both pro-inflammatory- and anti-inflammatory cytokines augments their
immuno-suppressive effects. Taken together, results from previous research support that
inhalational hypnotics are immuno-suppressive in mice as well as in humans.
Earlier findings also indicate other adverse effects of inhalational hypnotics that could be
related to immunological processes. For example, inhalational hypnotics seem to increase the
occurrence of cancer metastases. These adverse effects have not been found for propofol. In
contrast, propofol seem to inhibit tumor growth and reduce the tendency to induce
The research field of immuno-modulation from anesthetics was recently reviewed by Kurosawa
and Kato. They concluded, that "clinical anesthesiologists should select anesthetics and
choose anesthetic methods with careful consideration of the clinical situation and the
immune status of critically ill patients, in regard to long-term mortality, morbidity, and
the optimal prognosis". A key note is, that "many in vitro investigations have elucidated
the dose-dependent and time-dependent immunosuppressive effect of volatile (read
inhalational) anesthetics on various immune cells". It was stressed in another review by
Meiler, "that the perioperative process could be responsible for later adverse events", and
the necessity to "understand the underlying biology and immunology should be particularly
helpful in this pursuit". Thus, the choice of hypnotic may affect survival after cancer
surgery. More specifically, the combined effects of surgical stress and the burden of cancer
and perhaps other aggravating circumstances, such as high age and malnutrition, may play a
salient role in postoperative morbidity and mortality.
B) Genotoxicity Genotoxic agents may negatively affect patient's survival after cancer
surgery, as the connection between DNA damage and cancer development is well-known. The
potential genotoxicity from inhalational anesthetic agents in patients and in exposed staff
in operating rooms has been studied both in vitro and in vivo. A dose-response relationship
for inhalational agent exposure and DNA damage has been suggested. The techniques used, the
rate of sister chromatid exchange in lymphocytes and the alkaline comet assay, as indicators
of genotoxicity are well validated, and they are frequently used in other contexts.
Inhalational agents seem to be consistently genotoxic, whereas the less studied propofol
seems not to be so.
Project description: Retrospective, follow-up cohort study.
Patients We have in an administrative system demographic-, anesthetic-, and surgical data
logged by computer available for all patients exposed to anesthesia and surgery, dated from
January 1, 1998 to December 31, 2009. Data includes the choice of hypnotic. All approx. 4500
patients operated on for breast-, colorectal-, or skin cancer (malignant melanoma) have been
extracted from this register. Remaining demographic-, anesthetic-, and surgical data of
interest will be extracted from patient's individual paper files. A data base will be
constructed in the Statistical Program for Social Sciences, SPSS (Chicago, IL, USA).
Accessible outcome data are stored at the Regional Oncologic Center in Uppsala.
Bias, especially potential major confounders, control of
1. The risk of selection bias is an inherent major disadvantage with a retrospective
study. Here the risk is considered low due to the non-selective use of the hypnotics.
Misclassification errors will be corrected on an individual level and important
confounders identified at group level, c.f. down.
2. The distribution of different demographic characteristics between the two groups will
be controlled for by the thorough survey of every individual patient file. Data on type
and stage of the cancer, as well as different prognostic markers will also be included.
Any unequal distribution of such confounders will be identified. Data on alcohol
consumption will not be available.
3. Opioids also affect the immune system, and they will therefore theoretically constitute
to be major confounders. This holds true for morphine, which we never use
intra-operatively. Synthetic opioids, such as fentanyl, alfentanil, and remifentanil,
all used by us intra-operatively, have been proven not to suppress the immune response
like morphine does. On the contrary, the synthetic opioids may have positive effects in
this context. We use synthetic opioids for every surgical patient at our hospital, as
most other hospitals do worldwide. We have no reason to suspect an uneven
administration of postoperative morphine to the two study groups, but this has also to
be controlled for during data extraction.
4. Nitrous oxide, impairs the immune defense, and it also impairs DNA production by
inhibition of the vitamin B12 component of methionine synthetase. Nitrous oxide will be
a potential confounder for a fraction of our study population. It remains to be seen
during a thorough survey of our data, if the use of nitrous oxide was evenly
distributed between our two study groups?
5. Red blood cell transfusion may affect survival after cancer surgery. An
immuno-suppressive effect from the allogenic material is one suggested cause. We will
identify those patients having peri-operative red blood cell transfusions, control the
distribution between the groups, and include them in a subgroup analysis.
6. A preceding or subsequent anesthetic given in proximity to the index procedure
constitutes an important confounder and must therefore be identified. An executive
decision was made defining this time span of ± 1 year from the index operation.
Patients, who were anesthetized once or more than so besides the index operation within
this time frame, and patients who were anesthetized from the end of the time frame to
inclusion in the study will constitute separate subgroups. Accordingly, they will be
analyzed separately. We assume that 10-20% of patients had another general anesthetic
within the defined time interval. This assumption is based on a small pilot
investigation of 100 patients in the cohort.
7. Different adjuvant cancer therapies with potent toxicity and serious side-effects will
also be registered and controlled for from a confounding point of view.
Linking and matching of databases The administrative database, created as described above,
will be record-linked to the regional quality registers at the Regional Oncologic Center
(ROC) in Uppsala. The registers have been found to be >97% completeness compared to the
Swedish Cancer Register (SCR), to which reporting is mandatory and regulated by law. The SCR
holds diagnoses only and contains no clinical information. The ROC includes a quality
register for breast cancer, which was started in 1992. A register for colon cancer was
started in 1995, another for melanoma was started in 1996, and one for rectal cancer was
created in 1997. These registers contain information on mode of detection, histopathology,
stage at diagnosis, other prognostic markers and complementary treatment given. Hence,
complete oncologic- and outcome data will be available for all types of cancer included in
the study within the defined period of time. Data on type and stage of the cancer, as well
as different prognostic markers recorded in the oncologic registers will be extracted, as
well as data on recurrences of disease, vital status and date and cause of death. The unique
personal identification number will be replaced by unidentified serial numbers, to ensure
anonymity for each person after data securing.
Analysis The main endpoint will be a comparison of overall survival using time to event
analysis. Cumulative 1- and 5-year overall survival will be assessed using the Kaplan Meier
method and the estimates will be compared between patients given sevoflurane or propofol. In
a next step, Cox Proportional hazard models will be calculated to assess the risk of death
adjusted for potential effect modifiers and confounders. There will also be stratifications
for different types of tumors, cardio-pulmonary status, ASA-class etc. Hazard ratios (HR)
with 95 % confidence intervals (CI) will be presented for all models. Subgroup analyses will
be undertaken for patients having more than one anesthetic within the defined time frame; a)
given the same hypnotic as during the index procedure, b) given the other hypnotic at the
different occasion(s); and c) for patients having one or more anesthetics outside the
defined time frame.
Statistics A clinically relevant absolute difference in five-year-survival would be 5%, e.g.
a difference between 85% and 80%, the latter being an average one-year-survival for the
period of interest for breast cancer, which constitutes the largest group in the study. With
more than 3000 patients anesthetized with sevoflurane, and more than 1000 patients
anesthetized with propofol, we will have a 95% power to detect the difference with a P-value
of <0.05, i.e. there will be a good margin for an unanticipated loss of data or ditto
Ethics The project was accepted by the Regional Ethics Committee Jan 21, 2009 (2008/350).
The retrospective approach will not create ethical considerations.
Significance Undesired effects from anesthesia on survival have strong relevance for the
overall cancer treatment. Any suspicions of such a dramatic side-effect as potentially fatal
immuno-modulation or genotoxicity must be investigated.
If the hypotheses are confirmed the results from the present study will be used for a
large-scale randomized controlled trial (RCT) at multiple centers to compare the two agents,
and this in turn might result in a change in practice should the results be positive. If the
hypothesis is ruled out, the choice between inhalation and intravenous anesthesia will not
be influenced by oncological considerations, which is important to know, although it also
have to be confirmed in a RCT.
Observational Model: Cohort, Time Perspective: Retrospective
One year from index procedure
Mats Enlund, M.D., Ph.D.
Uppsala University, Centre for Clinical Research-Vasteras
Sweden: The National Board of Health and Welfare