Discussion:
In this study, we analyzed a retrospective cohort of AYA with
osteosarcoma treated at our center using a uniform non-HDMTX-based
protocol “OGS-12”, and Ewing’s sarcoma treated with our institutional
standard “EFT-2001” protocol. We formulated and validated prognostic
scores separately for metastatic and non-metastatic osteosarcoma and
Ewing’s sarcoma based on baseline clinical factors and tailored to a
unique population of patients treated in a resource constrained setting
with our in-house, low-cost, institutional standard protocols which have
previously demonstrated survival outcomes are comparable to those
published from Western countries [12-15].
For non-metastatic AYA sarcomas, features of high tumor burden including
higher baseline SAP, neurovascular bundle involvement and joint
involvement predicted poor EFS. This is consistent with published
literature - serum alkaline phosphatase and lactate dehydrogenase are
surrogates of osteoblastic activity and thus, elevated levels may
indicate increased disease aggressiveness [16]. Additionally,
vascular involvement noted radiologically in pretreatment MRI has been
found independently to be an risk factor for OS and EFS in patients with
Enneking IIB primary osteosarcoma involving extremities. [17] Poor
histological necrosis also predicted inferior EFS, which is well
described in existing literature.
Importantly, failure to complete treatment and prior treatment
independently predicted inferior EFS. This is particularly significant
in LMIC, with high treatment abandonment rates as well as failure to
complete treatment protocol due to various issues including financial
challenges, lack of education and motivation for treatment and logistic
issues relating to the need to travel long distances and stay far from
the patient’s hometown for the duration of treatment due to
unavailability of cancer care near the place of residences. Thus, prompt
referral to dedicated centres specializing in cancer care and awareness
of primary health care physicians to recognize bone sarcomas and avoid
inadvertent inappropriate medical or surgical intervention is of
paramount importance.
For Ewing’s sarcoma, tumor size >8cm was an additional
prognostic factor, consistent with reported literature. [18-19] In
the metastatic cohort, failure to complete treatment protocol, poor
histological necrosis, tumours not amenable to local treatment and
>10 metastases were also predictive of superior EFS on
multivariable analysis. determinants of survival. This underscores the
importance of multimodality clinics to identify tumors amenable to
curative treatment and avoid overtreatment for patients with extensive
disease. It has been previously observed that osteosarcoma presenting
only with lung metastases have better survival outcomes than metastases
at other sites [20].
The factors that emerged prognostically significant for our patients
were similar to those described in data from centres using HDMTX-based
protocols, indicating the wide applicability of our results. For
non-HDMTX regimens, particularly prevalent in LMICs, data on prognostic
factors are scarce with only few observational studies [21-23].
Histologic response to chemotherapy has been uniformly reported to be a
predictive factor in these studies. A study from Brazil has additionally
reported presence of metastases at baseline, primary tumor site and type
of surgery (amputation vs. limb sparing) as prognostically significant
[24]. Nevertheless, the lacunae for studies specifically targeting
bone sarcomas in LMICs remains apparent. Smaller sample sizes of
existing studies, observational nature, and non-uniformity of treatment
protocols makes generalizability of these results challenging. A
potential strategy to overcome these shortcomings include collaborative
efforts with multi-institution studies to enhance understanding of
osteosarcoma in LMICs. Importantly, several existing studies have
included patients with both non-metastatic and metastatic tumors, with
widely differing outcomes. We have addressed this limitation by
separately analyzed homogenously treated populations of metastatic and
non-metastatic osteosarcoma and Ewing’s sarcoma.
In our study, we developed separate prognostic models for non-metastatic
and metastatic osteosarcoma and Ewing’s sarcoma based on the prognostic
factors identified. We validated the risk score for non-metastatic
osteosarcoma, and demonstrated effective discriminative ability for
event-free survival between the three risk groups. For the other cohorts
(metastatic osteosarcoma, and non-metastatic and metastatic Ewing’s
sarcoma), we developed prognostic models depicted as nomograms, although
validation could not be separately performed due to the limited sample
size. Available prognostic models for risk stratification in
osteosarcoma involve variable treatment modalities, clinician
preferences, surgical expertise and chemotherapy protocols which
precludes generalizability of these models. [25-28]. For Ewing’s
sarcoma, existing models have largely utilized treatment-related factors
such choice of local treatment, factors that vary in different centres
due to individual clinician preference and expertise. [29-34] In
developing our prognostic models, we have analyzed homogenously treated
cohorts and reinforced the importance of baseline indicators of tumor
aggressiveness, chemosensitivity as indicated by histologic response to
chemotherapy, and general treatment-related factors unique to patients
in LMIC settings, such as failure to complete treatment protocol and
prior inadvertent treatment before referral to advanced oncology
centres.
The prognostic models we report are important and unique as they are
derived from a prospective analysis of homogenously treated AYA in a
LMIC setting. The specific challenges faced by this vulnerable subset of
patients has been stressed upon. The wide implications of our results
include healthcare policy-making, in the context of need for early
identification by appropriate awareness and training of primary care
physicians and timely referral to sarcoma reference centres, and the
importance of strategies to improve protocol completion by efforts such
as targeted nutritional intervention, extended growth factor support and
patient navigation facilities. Risk stratification of patients aids
selection of patients who would benefit from aggressive treatment as
well as those who may achieve comparable outcomes with less aggressive
modalities with improved quality of life. Traditionally, escalation of
treatment for patients with high-risk disease has been based on
histologic response after completion of neoadjuvant chemotherapy
[35-36]. Use of other important prognostic factors as described in
our study may be used to identify patients who may benefit from a less
intensive approach such as metronomic chemotherapy regimens and early
incorporation of palliative care, particularly in resource-constrained
settings [37-38].
The prognostic implication of common social challenges have been
highlighted, including prior inadvertent treatment and the importance of
compliance in the form of protocol completion. We did not analyze
socioeconomic strata due to unavailability of uniform data, however,
surrogate parameters including compliance as assessed by treatment
duration and protocol completion, as well as prior inadvertent treatment
were analyzed. Importantly, majority of our patients are constrained
financially and socially and depend on financial aid from various
government schemes for cancer treatment. Another study from India has
evaluated location of primary residence of the patient and the distance
of the residence from the hospital and not found prognostic significance
for osteosarcoma [23].