1
The Angiosarcoma Project: enabling genomic and clinical discoveries in a rare
cancer through patient-partnered research
Corrie A. Painter
1,2
£
, Esha Jain
1-3
£
, Brett N. Tomson
1,2
£
, Michael Dunphy
1,2
, Rachel E. Stoddard
1,2
,
Beena S. Thomas
1,2
, Alyssa L. Damon
1,2
, Shahrayz Shah
1,2
, Dewey Kim
1-3
, Jorge Gómez Tejeda
Zañudo
2,3
, Jason L. Hornick
4
, Yen-Lin Chen
5
, Priscilla Merriam
3,6
, Chandrajit P. Raut
6,7
, George D.
Demetri
3,6,8
, Brian A. Van Tine
9
, Eric S. Lander
1,2,10, 11
, Todd R. Golub
1,2,12
, Nikhil Wagle
1-3,13-14
*
Affiliations:
[1] Count Me In, Cambridge, MA.
[2] The Broad Institute of MIT and Harvard, Cambridge, MA.
[3] Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA.
[4] Department of Pathology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA.
[5] Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA.
[6] Center for Sarcoma and Bone Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA.
[7] Department of Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, MA.
[8] Ludwig Center at Harvard, Harvard Medical School, Boston, MA.
[9] Division of Oncology, Washington University School of Medicine in St. Louis, St. Louis, MO.
[10] Department of Biology, Massachusetts Institute of Technology, Cambridge, MA.
[11] Department of Systems Biology, Harvard Medical School, Boston, MA.
[12] Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA.
[13] Center for Cancer Precision Medicine, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA.
[14] Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA.
£
These authors contributed equally
*Address correspondence to:
Nikhil Wagle, MD
Department of Medical Oncology
Dana-Farber Cancer Institute
450 Brookline Ave, Dana 820A
Boston, MA 02215
Phone: 617-632-6419
E-mail: nikhil_wagle@dfci.harvard.edu
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted August 26, 2019. ; https://doi.org/10.1101/741744doi: bioRxiv preprint
2
ABSTRACT
Despite collectively accounting for 25% of tumors in U.S. adults, rare cancers have
significant unmet clinical needs as they are difficult to study due to low incidence and
geographically dispersed patient populations. We sought to assess whether a patient-
partnered research approach using online engagement can overcome these challenges
and accelerate scientific discovery in rare cancers, focusing on angiosarcoma (AS), an
exceedingly rare sarcoma with a dismal prognosis and an annual U.S. incidence of 300
cases. Here, we describe the development of the Angiosarcoma Project (ASCproject), an
initiative enabling patients across the U.S. and Canada to remotely share their clinical
information and biospecimens for research. The project generates and publicly releases
clinically annotated genomic data on tumor and germline specimens on an ongoing
basis. Over 18 months, 338 AS patients registered for the ASCproject, comprising a
significant fraction of all patients. Whole exome sequencing of 47 AS tumors revealed
several recurrently mutated genes, including KDR, TP53, and PIK3CA. Activating
mutations in PIK3CA were observed nearly exclusively in primary breast AS, suggesting
a therapeutic rationale in these patients. AS of the head, neck, face, and scalp (HNFS)
was associated with high tumor mutation burden and a dominant mutational signature of
UV light exposure, suggesting that UV damage may be a causative factor in HNFS AS
and that this AS subset might be amenable to immune checkpoint inhibitor therapy.
Medical record review revealed two patients with HNFS AS received off-label treatment
with anti-PD-1 therapy and experienced exceptional responses, highlighting immune
checkpoint inhibition as a therapeutic avenue for HNFS AS. This patient-partnered
approach has catalyzed an opportunity to discover the etiology and potential therapies
for AS patients. Collectively, this proof of concept study demonstrates that empowering
patients to directly participate in research can overcome barriers in rare diseases and
enable biological and clinical discoveries.
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted August 26, 2019. ; https://doi.org/10.1101/741744doi: bioRxiv preprint
3
Due to low incidence, rare cancer patients are often treated at disparate institutions distributed
across the country, ranging from tertiary medical centers to community hospitals. This poses
barriers to large-scale scientific studies urgently needed to understand the biology of rare
cancers and develop better treatments
1
. We hypothesized that the challenges of rare cancer
research could be addressed by engaging patients directly and empowering them to share their
samples, their data, and their experiences. In principle, a patient-partnered approach that
harnesses the power of social media and patient networks and enables patients to remotely
participate irrespective of geography could overcome the barriers of low patient numbers seen
at any single institution and aggregate a significant number of rare cancer patients from
numerous institutions in a unified clinicogenomic study, thereby rapidly yielding discoveries.
We aimed to test this hypothesis in angiosarcoma, a disease that represents just 1-2% of soft
tissue sarcomas, which in turn, comprise less than 1% of adult malignancies
2,3
. The prognosis
for AS is poor, with a reported 5-year disease-specific survival of 38%
2
. Although there have
been small genomic studies of AS to date
4-10
, the majority of AS have no known genomic,
environmental, or iatrogenic etiology, and effective therapies for most AS patients are lacking.
Working closely with patients and patient advocates, we developed a website (ASCproject.org),
which allows AS patients living anywhere in the United States and Canada to register for the
ASCproject (Figure 1A). Patients in the angiosarcoma community were deeply involved in all
aspects of the project design, implementation, testing, and refinement, including all elements of
the study website from images to consent language. AS patients joined the ASCproject rapidly
after launch, with 120 patients registering in the first month and a total of 338 patients
registering within 18 months (Figure 1B). This represents not only a significant proportion of
people living with this disease in the U.S., but also a substantially increased pace of enrollment
compared to previous efforts (with the largest previous AS study having collected clinical data
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted August 26, 2019. ; https://doi.org/10.1101/741744doi: bioRxiv preprint
4
from 222 patients treated over 14 years
2
). Online consent for the ASCproject allowed for
acquisition of medical records and biological samples (tumor, saliva, and blood), analysis of
whole exome sequencing (WES) on tumor and germline DNA, and sharing of de-identified
patient-reported, clinical, and genomic data via public databases (Figure 1C-D). Patients
continued to be engaged throughout the ASCproject and were regularly provided study updates
(Figure 1C).
Although the study is ongoing, the following analyses were conducted with the 227 patients who
had fully consented as of September 30, 2018 (Figure 1D). These 227 patients received care for
AS at 340 different clinical institutions, including 289 institutions that were reported only once by
any given participant (Supplemental Figure 1) - demonstrating the importance of online
platforms to overcome the geographic isolation that has traditionally inhibited large-scale studies
in rare cancer patients.
Patients self-reported demographic information, sites of primary AS, as well as other AS and
prior cancer information through an intake survey (Figure 2; Supplemental Figures 2 and 3;
Supplemental Tables 1 and 2). Patients joining the ASCproject spanned newly diagnosed
patients to long-term survivors, with the elapsed time between primary AS diagnosis and
ASCproject enrollment ranging from 5 days to 41 years (Figure 2B).
We were able to rapidly acquire medical records and tumor samples from geographically
dispersed patients and institutions (Supplemental Figure 4). We performed WES on 70 obtained
tumor samples. Forty-seven samples from 36 patients were used for subsequent genomic
analysis after assessment of sufficient tumor purity (³10%) and confirmation as angiosarcoma
by centralized pathology review (Supplemental Figure 4). Apart from these considerations, there
were no additional selection criteria for these 47 samples. Characteristics of the 36 sequenced
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted August 26, 2019. ; https://doi.org/10.1101/741744doi: bioRxiv preprint
5
patients are shown in Supplemental Figure 5. Abstraction of medical record data (Supplemental
Table 3) and histological evaluation were used to classify these tumors into 8 subclassifications
of AS (see Supplemental Methods). To our knowledge, this is the largest reported cohort of AS
samples that have undergone WES.
We found 30 genes recurrently altered in these 47 samples (determined by somatic alteration
frequency, see Methods). This includes genes previously reported as altered in AS
6,7,10-12
, as
well as several genes that have not been previously reported to be mutated in AS, such as
PIK3CA, GRIN2A, and NOTCH2 (Figure 3A). Two genes were mutated at a rate significantly
higher than expected by chance given background mutational processes (as identified by
MutSig2CV
13
; see Supplemental Methods): TP53 (25%; 9/36 patients) and KDR (22%; 8/36
patients) (Supplemental Figure 6). Moreover, mutations in these two genes were also mutually
exclusive (p-value=0.02), with 89% (8/9) of KDR missense mutations being observed in primary
breast AS samples and 82% (9/11) of TP53 missense mutations detected in AS samples that
were not primary breast (Figure 3A; Supplemental Figure 6).
PIK3CA was one of the most frequently mutated genes in this cohort (21%; 10/47 samples)
(Figure 3A). Although alterations in the PI3K pathway have been identified in a previous AS
study
14
, mutations in the PIK3CA gene itself have not been previously reported in AS to our
knowledge. Nine out of the ten PIK3CA alterations were found in primary breast AS samples,
and this AS subclassification was significantly enriched for PIK3CA mutations compared to
other subclassifications (9/18 primary breast AS samples versus 1/29 AS samples that were not
primary breast; p-value=0.0003) (Figure 3A).
Intriguingly, none of the 8 unique PIK3CA mutations we observed were at the canonical PIK3CA
hotspot residues E545 or H1047
15
(Figure 3B). Instead, these PIK3CA mutations were located
certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was notthis version posted August 26, 2019. ; https://doi.org/10.1101/741744doi: bioRxiv preprint