Cancer Therapy Vol 1, 245-256, 2003.
Activity of chemotherapy and
immunotherapy on malignant mesothelioma: a systematic review of the literature
with meta-analysis
Berghmans T1, Lafitte JJ3, Mascaux C1, Meert AP1,
Paesmans M2 and Sculier JP1
1Service de MŽdecine Interne et Laboratoire
dÕInvestigation Clinique et dÕOncologie ExpŽrimentale and 2Data Centre, Institut Jules Bordet, Centre des Tumeurs
de lÕUniversitŽ Libre de Bruxelles, Belgium and 3Department of
Pneumology, CHU Calmette, Lille, France
__________________________________________________________________________________
*Correspondence: Dr Thierry Berghmans, Institut Jules Bordet, Rue HŽger-Bordet, 1, 1000
Bruxelles, Belgium; Tel: 322/541.31.11; Fax: 322/534.37.56; e-mail: thierry.berghmans@bordet.be
Key Words: chemotherapy,
immunotherapy, mesothelioma, quality score, systematic review, meta-analysis
Abbreviations: cisplatin CDDP;
vinblastine VBL; interferon INF; interleukine 2 IL-2; mitomycin C MMC;
doxorubicin Doxo; ifosfamide Ifo cyclophosphamide CPA; bleomycin Bleo;
hyaluronidase Hyal; global quality score QS; internal validity IV; external
validity EV; response rate RR; confidence interval CI
Summary
Malignant
mesothelioma is a tumour with increasing incidence, for which treatment remains
debatable. Different chemotherapy regimens have been tested in phase II
studies. The aim of the present report is to update the results of a previous
meta-analysis to identify new chemotherapy regimens which could be selected for
future randomised trials. Ninety-five articles corresponding to 100 treatment
arms, published between 1983 and 2003, were eligible for the analysis. A
qualitative evaluation was performed using the ELCWP methodological quality
scale. No statistically significant difference in term of methodological score
was found between the positive and potentially positive (upper limit of the 95%
confidence interval (CI) for the response rate (RR) > 20%) and negative
studies (median: 57.3% versus 56.7%; p = 0.68), allowing us to perform an
aggregation of the results of the different studies. We found that the most
active combination regimens, in terms of response rate are cisplatin plus doxorubicin
(RR 28.5%), gemcitabine (RR 29.8%) or etoposide (RR 27.1%). When single agent
therapy is considered, cisplatin seems to be the most active single agent (RR
17.0%). No other endpoints such as survival, toxicity or quality of life could
be meta-analysed, due to the lack of data in the publications. Cisplatin plus
doxorubicin, gemcitabine or etoposide appear as the most active regimens in the
treatment of malignant mesothelioma. These results must be interpreted in the
context of the recently published phase III trial demonstrating a significant
survival advantage of cisplatin plus pemetrexed on cisplatin monotherapy.
The incidence of
malignant mesothelioma is expected to rise in the next few years, due to
increased asbestos exposure during the last decades (Driscoll et al, 1993). Few treatments have
demonstrated activity against this disease. The beneficial survival impact of
surgery, pleurectomy or extrapleural pneumonectomy, has never been proven in
randomised trial. Only a minority of the patients are eligible for these
treatments (Sugarbaker et al, 1991), whose mortality and
morbidity can be considerable. The efficacy of radiotherapy is not proven (Ong and Vogelzang, 1996).
A majority of the
patients are medically treated. Some chemotherapeutic agents such as cisplatin
or doxorubicin are considered as potentially useful. In a previous systematic
review of the literature with meta-analysis, we found that cisplatin-based or
cisplatin and doxorubicin-based chemotherapy were the most active regimens with
respective response rates of 23.2% and 28.3% (Berghmans et al, 2002). Some new agents have been
tested the last two years in phase II studies, including gemcitabine (Nowak et al, 2002;van Haarst et
al, 2002), taxans (Vorobiof et al, 2002), raltitrexed (Baas et al, 2003;Fizazi et al, 2003;Maisano et al,
2001) or oxaliplatin (Fizazi et al, 2003; Maisano
et al, 2001) and the first
randomised phase III trial comparing two chemotherapeutic regimens has been
published (Vogelzang et al, 2003).
In the present report, we update our previous
systematic review on the activity of chemotherapy and immunotherapy in
malignant mesothelioma, in order to identify new cisplatin-based chemotherapy
regimens which could be selected for future randomised trials.
The search for
prospective published trials relative to the treatment of malignant
mesothelioma of pleural or peritoneal origin was performed by consulting the
Medline, Health Star and National Cancer Institutes electronic data bases and
completed by references found in the papers, in textbooks, in reviews and those
known by the investigators.
The criteria of
eligibility of the articles were the following: to focus only on patients with
malignant mesothelioma; to be related to the study of single or combined
cytotoxic and/or immunomodulatory agents, administered by systemic or local
routes; to be published in English, French or Dutch languages between 1965 and
January 2003; to be a prospective single or randomised phase II or phase III
trial with a minimum of 14 patients included. If less than 14 patients were
included in a prospective phase II trial, the study could be considered as
eligible if at least one objective response was observed when targeting a
response rate of 20%, according to the GehanÕs design for phase II studies (Gehan EA, 1961). Abstracts were excluded from
this analysis because of insufficient data to apply the scoring system and to
evaluate the methodological quality of the trial.
The methodological
qualitative evaluation was performed by a team of 5 medical doctors and 1
biostatistician. Consensual agreement on the scores attributed to each item for
each trial was obtained by regular meetings The study quality was assessed
according to the information provided in the publication, using the previously
published ELCWP quality scale (Berghmans et al, 2002). All items were grouped in ten
categories and a global quality score as well as two subscores assessing the
internal and external validities of the studies were calculated.
For each article,
the numbers of eligible patients were recorded by applying the criteria used in
ELCWP trials (Sculier et al, 1996) considering toxic death, early
death due to cancer or treatment discontinuation due to toxicity as treatment
failures. We assumed that a chemotherapeutic agent had a clinical potentially
useful activity in a trial if its objective response rate was at least 20%. We
considered that a study was negative if the upper limit of the 95% confidence
interval (CI) of the response rate was £ 20%. It was
considered as positive if the lower limit of the 95% CI was > 20% and as not
conclusive but potentially positive if the upper limit of the 95% CI was >
20% but the lower limit < 20%.
Descriptive
summary statistics were calculated in each category for the distributions of
the scores (internal validity, external validity and global score). Normality
of the distribution of continuous variables was assessed by a
Kolmogorov-Smirnov test. If the distribution was normal, the distribution of
these continuous variables according to the levels of a categorical variable
were compared by using parametric tests (ANOVA or Student). Otherwise, non
parametric tests (Wilcoxon or Kruskal-Wallis) were applied. Relationship
between the scores and other continuous variables was assessed by the
calculation of Pearson or Spearman correlation coefficients, according to the
normality of the distributions of the continuous variables. Confidence
intervals for the response rate to the chemotherapeutic regimen were, for
consistency, recalculated using the exact binomial distribution. Proportions
were compared with chi square tests for homogeneity. All reported p values are
two-tailed.
A. Phase II studies
1.
Trials
characteristics
Ninety-five
articles, published between 1983 and 2003, met our selection criteria and were
eligible for the analysis. Out of these 95 eligible studies, 92 were single arm
phase II trials and 3 were randomised phase II trials. For the purpose of this
review, each arm of the randomised studies was assessed as an independent
trial. In 2 papers, 2 separate phase II trials were reported in the same
publication. Thus, 100 ÇarmsÈ, each considered as an independent study, were
analysed. They will be further called ÇstudiesÈ.
In a first
analysis, the studies were separated in 4 groups according to the treatment
regimen (Table 1). Group 1 (n = 22) corresponded to the trials testing
cisplatin but not doxorubicin. Group 2 (n = 9) was composed of the trials
investigating doxorubicin without cisplatin. Seven studies, assessing a
combination including both cisplatin and doxorubicin, formed Group 3. The last
62 trials, with regimens without cisplatin or doxorubicin, were included in
Group 4. Sixty-two studies (62%) used a single agent regimen. The chemotherapy
mainly consisted in platinum (n = 32) and/or anthracycline (n = 27)
derivatives. In 20 studies, an immunomodulatory agent was used either alone (n
= 8) or combined with cisplatin (n = 8) or with other agents (n = 4). Other
tested agents and regimens are detailed in Table 1 (4,12-93).
Among the 100
eligible studies, 37 were negative, 5 positive and 58 potentially positive in
term of antitumoral response, as defined above. For the purpose of the
analysis, the potentially positive trials were pooled with the true positive
ones and this whole group will be further named as the Çpositive trialsÈ. The
total number of patients assessable for response and incorporated in the 100
studies was 2727.
Other endpoints
have been considered but the lack of data or their presentation precluded to
perform meaningful quantitative aggregation. Survival rates were not reported
in 18 of the 100 analysed arms. Thirteen studies reported on symptoms
evaluation (n = 10) or quality of life (n = 3). Toxicity was only fully
described in 46 arms; partial information was available in 44; it was not analysed in 10.
2.
Methodological assessment
The results of the
qualitative methodological evaluation for each trial are given in Table 1. The overall mean
and median scores attributed per score subscales are described in Table 2. No statistically
significant difference in term of methodological score was found between the
positive and negative trials whatever global (median: 57.3% versus 56.7%; p =
0.68), internal (45.8% versus 43.3%; p = 0.58) or external (68.9% versus 71.5%;
p = 0.87) validity scores were considered. No statistically significant
difference was observed between the 4 groups according to the type of
therapeutic regimen (p = 0.42) (Table 3). A significant difference
was found according to the method of tumour response assessment (Table 4): studies using
radiological techniques as a part or as the whole of the evaluation had better
scores than the others (p < 0.0003) but were also significantly more recent
(p < 0.001).
Table 1. Treatment regimen,
quality scores and response rates with 95% confidence interval (ELCWP)
according to treatment group for assessable patients
|
Schedule |
n pts |
QS (%) |
IV (/50) |
EV (/50) |
RR (%) |
95%CI |
|
Group 1 (n = 22)
Cisplatin without doxorubicin containing regimens |
|
|
|
|
|
|
|
CDDP (Planting et al,
1994) |
14 |
56.5 |
19.2 |
37.3 |
35.7 |
7-64.4 |
|
CDDP (Markman et al, 1986) |
21 |
30.9 |
5.0 |
25.9 |
14.3 |
0-31.6 |
|
CDDP (Zidar et al, 1988) |
35 |
46.1 |
12.5 |
33.6 |
14.3 |
1.3-27.3 |
|
CDDP (Mintzer et al, 1985) |
24 |
36.9 |
12.1 |