Keywords targeted therapy - India - low- and middle-income countries - metastatic colorectal
cancer - overall survival - resource constrained
Introduction
Metastatic colorectal cancers (mCRC) are a diverse group of cancers with respect to
biomarkers, targetable mutations, and treatment patterns. The emergence of RAS/Raf
testing as a biomarker for selection of monoclonal antibodies (moAbs; anti-EGFR agents
and anti-VEGF agents) and the subsequent use of moAbs has improved upon the survival
outcomes seen in mCRCs with systemic chemotherapy alone.[1 ]
[2 ] The median overall survival (mOS) has risen from approximately 14 to 15 months with
chemotherapy alone a decade ago to approximately 30 months in certain subgroups, using
a combination of chemotherapy and targeted therapy.[3 ] Additional factors which have improved survival in the last decade include optimization
of treatment strategies in the first line (1L), increased use of curative intent treatment
strategies in oligometastatic mCRC with liver and lung limited disease, as well as
third line (3L) options such as regorafenib and TAS 102.
However, limited feasibility with regard to the use of potentially expensive moAbs
in resource-constrained settings may hamper the applicability of survival improvements
seen in trials to real-world clinical practice. Such logistic and economic constraints
in terms of reduced usage have been recognized with specific regard to bevacizumab,
cetuximab and panitumumab in mCRC across centers in predominantly Eastern Europe and
to some extent in Western Europe as well. The advent of pembrolizumab and nivolumab
in mCRC may further widen such differences between trial data and use in clinical
practice. Biosimilars, which are similar versions of licensed biologicals, may offset
some of these differences, especially in countries like India where they are routinely
available.[4 ]
With the above reference points in the background, the authors conducted a retrospective
analysis of patients with mCRC who were treated in the Tata Memorial Hospital, Mumbai.
The aims of the study were to evaluate demographics, treatment patterns and outcomes
in patients with mCRC, as well as evaluate receipt of moAbs. Additionally, an attempt
was made to throw light on the costs of treatment for mCRC in India and potentially
link these variables with how it affects outcomes.
Materials and Methods
The study is a retrospective analysis of mCRC patients who were evaluated during the
period of January 2013 to August 2017 in the Department of GI Medical Oncology at
Tata Memorial Hospital. Data was obtained from a prospectively maintained metastatic
CRC database. Decisions regarding metastatic nature of disease was made by a dedicated
gastrointestinal (GI) multidisciplinary joint clinic (MDJC).
Patients satisfying all the following criteria were included in the analysis:
Histologically proven colorectal adenocarcinoma, either by cytology or biopsy.
Definitive evidence of metastatic disease, as per scans and physical examination.
Administered at least one cycle of chemotherapy at our hospital or at least one follow-up
after starting treatment.
After MDJC, patients were evaluated for fitness for chemotherapy by the Department
of Medical Oncology and were then offered chemotherapy with targeted therapy based
on feasibility. All RAS and BRAF testing were only offered to patients who were financially
and logistically feasible for receipt of moAbs, based on discussion with the treating
medical oncologist. Baseline demographic details, including comorbidities, prior treatment
history, and therapeutic options used, were recorded. As per standard institution
criteria, baseline disease evaluation included, at minimum, a carcinoembryonic antigen
(CEA) test, biopsy (primary or metastatic site), contrast-enhanced CT scan of the
thorax, abdomen and pelvis or, rarely, a fluorodeoxyglucose (FDG) positron emission
tomography (PET) CT scan.
The investigators (AR and VO) drew up an imaginary treatment-outcomes and cost scenario
for a patient with full access to the complete sequence of treatment (patient A) for
mCRC as opposed to a patient with financial constraints to treatment (patient B) ([Fig. 1 ]). This construct was made, based on available survival data from randomized trials
and an estimate of treatment alone costs (without consideration of logistic costs,
costs of treatment-related investigations and treatment-related adverse events as
per 2017 figures) prior to data analysis of the study cohort. The investigators hypothesized
that median OS of the entire study cohort would range between 16 to 18 months, that
is, similar to imaginary patient B.
Fig. 1 Imaginary comparative treatment scenario—individualized treatment in an Indian patient
with complete access to treatment options (A ) versus nontrial Indian patient with limited access to therapy (B ).
Outcome Variables
Toxicity assessment was done at every patient visit and recorded as per National Cancer
Institute (NCI) – Common Terminology Criteria for Adverse (CTCAE) version 4.0. Response
to treatment was evaluated clinically on every visit with contrast-enhanced CT (CECT)
scan after three to four cycles of chemotherapy or earlier as per physician decision.
Prognostic factors evaluated included age (< 40 years vs. ≥ 40 years), site of primary
(right-sided vs. left-sided cancers), signet ring histology (presence versus absence),
mucinous histology (presence vs. absence), CEA levels at baseline (raised vs. within
normal limits), and resection of primary.
Progression-free survival (PFS) was calculated from date of diagnosis to date of progression,
cessation of chemotherapy due to adverse events, loss to follow-up, and withdrawal
from therapy or death (in case of no documented progression). PFS1 was calculated
for first-line chemotherapy and PFS2 for second line chemotherapy. OS was calculated
from date of diagnosis to date of death or loss to follow-up.
Clinical Data Collection and Statistics
For the purposes of this study, demographic data and baseline clinical data were collected
retrospectively from GI Medical Oncology Information System and electronic medical
record system. All data was entered in SPSS software version 21 and used for analysis.
Descriptive statistics, including median, frequency and percentage for categorical
variables, was used to describe age, gender distribution, treatment, and response
to treatment. Median PFS1, PFS2, and OS were calculated using Kaplan–Meier estimates,
while log rank test was used for univariate comparisons. Multivariate analysis by
Cox regression method for prognostic factors was done irrespective of results of univariate
analysis.
Results
Baseline Characteristics
A total of 403 patient's data satisfied the inclusion criteria, and they were available
for analysis. The median age of the cohort was 48 years (range: 17–86 years). Majority
of the cancers were rectal primaries (63.3%), with signet ring histology seen in 82
patients (20.3%). Other characteristics are detailed in [Table 1 ].
Table 1
Baseline demographic and clinical characteristics (n =403)
Characteristic
Number (percentage where applicable)
Median age (years)
48 (range:17–86)
• Age<40 years
129 (32)
• Age ≥ 40 years
274 (68)
Gender
• Female
155 (38.5)
• Male
248 (61.5)
Site of disease
• Left sided (nonrectal)
67 (16.6)
• Rectal
255 (63.3)
• Right-sided
63 (15.6)
• Transverse
17 (4.2)
• Epicenter not identified
01 (0.2)
Histopathology
• Poorly differentiated
113 (28)
• Well-differentiated/moderately differentiated
219 (54.3)
• Adenocarcinoma, not specified
71 (17.6)
Mucinous histology
• Yes
72 (17.9)
• No
331 (82.1)
Signet ring histology
• Yes
82 (20.3)
• No
321 (79.7)
Baseline CEA status
• CEA>ULN
246 (61)
• CEA ≤ ULN
67 (16.6)
• CEA not available
90 (22.3)
Disease status at baseline
• Baseline metastatic
183 (45.4)
• Recurrent metastatic
220 (54.6)
Prior adjuvant/neoadjuvant chemotherapy
• Yes
200 (49.7)
• No
203 (50.3)
Sites of metastases
• Liver
190 (47.1)
• Lung
127 (31.5)
• Peritoneal
133 (33)
• Nonregional nodes
145 (36)
• Osseous
32 (7.9)
• Krukenberg's
21 (5.2)
> 1 site of metastases
249 (61.8)
Abbreviations: CEA, carcinoembryonic antigen; ULN, upper limit of normal.
Characteristics of Chemotherapy ([Table 2 ])
Table 2
Characteristics of systemic chemotherapy
Characteristics
Number (percentage where applicable)
CT1
403 (100)
• mCAPOX
215 (53.3)
• FOLFIRINOX
17 (4.2)
• mFOLFIRI
85 (21.1)
• mFOLFOX-7
48 (11.9)
• Capecitabine monotherapy
13 (3.2)
• 5 FU/LV monotherapy
17 (4.2)
• Others
12 (2.9)
CT2
219 (54.3)
• mFOLFIRI
84 (38.4)
• CAPIRI
33 (15.1)
• mCAPOX/mFOLFOX-7
47 (21.5)
• mFOLFIRINOX
02 (0.9)
• Single agent irinotecan
10 (4.6)
• Capecitabine monotherapy
20 (9.1)
• 5 FU/LV monotherapy
03 (1.4)
• Metronomic chemotherapy
16 (7.3)
• Single agent cetuximab
01 (0.4)
• Regorafenib
01 (0.4)
CT3
84 (20.3)
• mCAPOX/mFOLFOX-7
18 (21.4)
• mFOLFIRI/CAPIRI
29 (34.5)
• Single agent irinotecan
03 (3.6)
• Metronomic chemotherapy
17 (20.2)
• Single agent monoclonal
01 (1.1)
• Capecitabine monotherapy
06 (7.1)
• Regorafenib
06 (7.1)
• Tegafur/Uracil
04 (4.8)
Patients receiving fourth-line therapy
31 (7.7)
Abbreviations: CT1, first-line chemotherapy; CT2, second-line chemotherapy; CT3, third-line
chemotherapy.
The most common first-line regimen (CT1) used was modified capecitabine-oxaliplatin
(CAPOX) in 215 patients (53.3%), while 85 patients (21.1%) received modified 5-fluorouracil/leucovorin-irinotecan
(mFOLFIRI). Of the 403 patients in total, 219 (54.3%) patients received second-line
therapy (CT2). The most common second-line regimens (CT2) used were irinotecan-based,
mFOLFIRI in 84 patients (38.3%) and CAPIRI (capecitabine-irinotecan) in 33 patients
(15.1%), followed by oxaliplatin-based regimens in 47 patients (21.5%). Third-line
treatment (CT3) was given in 84 patients of the entire cohort (20.3%; n =403). Metronomic chemotherapy (oral low dose capecitabine and cyclophosphamide) was
used in 38 patients (9.4%).
Characteristics of Targeted Therapy ([Table 3 ])
Table 3
Characteristics of targeted therapy
Characteristics
Number (percentage where applicable)
Use of moAbs with CT1 (n =403)
48 (13.4)
• Bevacizumab
25
• Cetuximab
23
Use of moAbs with CT2 (n =219)
34 (15.5)
• Bevacizumab alone
14
• Aflibercept
06
• Cetuximab
13
• Panitumumab
01
Use of moAbs with CT3 (n =84)
14 (16.7)
• Bevacizumab
06
• Cetuximab
06
• Panitumumab
02
Abbreviation: moAbs, monoclonal antibodies.
A total of 48 patients (13.4%) received moAbs with chemotherapy as part of CT1. Bevacizumab
was used in 25 patients, while cetuximab was administered in 23 patients. As many
as 34 patients (15.5%; n =219) received moAbs along with CT2, with bevacizumab being commonly used.
Survival Outcomes and Prognostic Factors ([Table 4 ])
Table 4
Factors affecting OS
Characteristic
OS (months)
p -Value (univariate analysis)
p -Value (multivariate analysis)
Hazard ratio (95% confidence interval)
Age
•<40 years
• ≥ 40 years
12.58
19.48
0.001
0.07
0.80 (0.623–1.019)
Site of primary
• Left-sided
• Right-sided
17.74
17.64
0.541
–
–
Mucinous histology
• Present
• Absent
20.38
16.30
0.449
–
–
Signet ring histology
• Present
• Absent
11.17
19.65
< 0.001
< 0.001
1.69 (1.282–2.229)
CEA levels at baseline
• Raised
• WNL
15.28
25.00
0.007
0.017
0.66 (0.472–0.929)
Resection of primary
• Yes
• No
23.95
9.70
< 0.001
< 0.001
3.075 (2.400–3.939)
Abbreviations: CEA, carcinoembryonic antigen; OS, Overall survival; WNL, Within normal
limits.
The median follow-up for the entire cohort was 39.49 months (range: 7.36–73.95). Median
PFS on CT1 (PFS1) was 10.91 months (95% confidence interval [CI]: 9.29–12.52), while
median PFS on CT2 (PFS2) was 7.39 months (95% CI: 6.65–8.14). A total of 303 patients
had died as of cut-off date for analysis, with median OS for the entire cohort being
17.61 months (95% CI: 15.48–19.74) ([Fig. 2 ]).
Fig. 2 Kaplan–Meier curve for overall survival (OS) of entire cohort.
Of the prognostic factors evaluated, the presence of signet ring histology (11.17
months Vs. 19.65 months; p <0.001) and raised CEA at baseline (15.28 months. versus 25 months.; p =0.017) predicted inferior OS, while patients with a resected primary had a superior
OS (23.95 months vs. 9.70 months.; p <0.001) on multivariate analysis. Younger age, while significant on univariate analysis,
did not predict inferior OS on multivariate analysis (p =0.07).
Discussion
The management of mCRC has grown remarkably over the last decade, be it with respect
to validated biomarkers, treatment options, or improved survival of patients. Current
median OS rates that are oft quoted are in the range of 24 to 36 months, especially
in trials which have examined chemotherapy-moAB combinations. However, differences
in patient selection between trials and real-world scenarios as well the added lack
of access to potentially newer treatment options hamper survival outcomes in low-
and middle-income countries (LMICS) with respect to moAbs.[5 ]
[6 ]
Our retrospective analysis of 403 patients with advanced CRC is the first such dataset
from India and is a reflection of treatment patterns and outcomes in a resource-constrained
scenario. A significant proportion of young patients (< 40 years—32%) and rectal cancers
(63.3%) and higher percentage of signet ring cancers (20.3%) were seen in the current
cohort and such trends have been shown previously from our institute.[7 ]
[8 ] The presence of an increasing proportion of younger patients having rectal cancer
is of significant importance, as there is growing evidence that they likely constitute
a biologically and clinically distinct cohort, who do not maximally benefit from current
treatment paradigm.[9 ] We have used a cut-off of 40 years in view of the lower median age of the entire
cohort.
The majority of patients were started on modified capecitabine-oxaliplatin (mCAPOX)
(capecitabine 2000mg/m2 instead of 2500mg/m2 ) (53.3%; n =403) as first-line chemotherapy (CT1) as opposed to mFOLFOX, as this avoids the use
of a central line without loss of efficacy. Only 54.3% of patients proceeded to CT2
post CT1, and this is at variance from patients who are able to receive CT2 in clinical
trials as well as real-world data (62–74%). A high-baseline disease burden (> 1 site
of metastases—61.8%) and possibly poor Eastern Cooperative Oncology Group performance
status (ECOG PS) with increased disease burden postprogression on CT1 may contribute
to lesser patients receiving CT2.
Two other treatment-related factors are significant in the current cohort—an increased
prevalence of a resected primary (69.7%; n =403) and limited use of targeted therapy with CT1 (13.4%) as well as CT2 (15.5%).
The increased percentage of a resected primary is because of the study cohort including
a high proportion (54.6%) with recurrent metastatic disease ([Supplementary Table S1 ], available online only). Available retrospective evidence and metanalysis have suggested
that carefully selected patients with mCRC may have survival benefit with resection
of the primary (and possible resection of metastatic sites).[10 ] Despite the findings in retrospective studies, including the current one, resection
of the primary should only be considered in clearly defined situations and not as
routine/standard in mCRC.
The limited administration of targeted therapy in the entire cohort is not unexpected.
In a LMICS country with a per capita of approximately US$ 1700 (2017 data) like India,
a majority of patients will only be able to afford CT1 and CT2 without targeted therapy,
and this would cost approximately INR 200,000 (US$ 2,800), assuming generic chemotherapeutic
agents are used ([Fig. 1 ]—patient B). The proportion of patients feasible for the entire gamut of treatment
options (similar to patient A) in terms of financial feasibility would be very few,
as per our institution data. A complete cost-effectiveness analysis using Markov models
would be required to evaluate the actual cost-benefit ratio of treatment sequencing
in mCRC in the Indian scenario, and this is beyond the scope of the current study.
The imaginary construct ([Fig. 1 ]) also was accurate in its prediction of median OS (predicted range: 16–18 months.)
for the study cohort. With a median follow-up of 39.49 months, the median OS was 17.61
months (95% CI: 15.48–19.74) and is reflective of patients receiving predominantly
two lines of chemotherapy (without moAbs) followed by supportive care.
An important subtext that emerges from this study is the importance of being able
to offer available treatment options to patients with mCRC. Just as the use of HER2
directed therapy has dramatically changed the outcomes in metastatic breast cancer,
so too has the use of anti-EGFR and anti-VEGF therapy improved outcomes in mCRC. The
use of targeted therapy and beyond extends OS to approximately 12 to 15 months over
chemotherapy alone. Such therapy-related factors should also be considered when governmental
funding of treatment is being planned and accounted.
Our study, while being retrospective in design and a single institution datum, has
the strength of being representative of how patients with mCRC are treated in the
LMICS scenario, where patient selection and availability of treatment options are
markedly different when compared with trial patients. However, multiple caveats in
the current study need to be acknowledged. In the current era, testing for all RAS
and BRAF in mCRC patients is almost mandatory, and we have not provided any information
on the same. This is primarily because of our policy of offering this test to patients
who are financially feasible for targeted therapy and not otherwise. We have attempted
to highlight financial constraints faced by patients in the Indian scenario in terms
of crude comparative absolute costs for treatment without using appropriate health
economic modelling approaches. Such an approach implies the need for a more systematic
evaluation of the cost-effectiveness of all aspects of systemic therapy for mCRC in
the Indian context.
