The leukocyte complexity and mutational landscape of periampullary adenocarcinoma. Morphology matters.

Background: Periampullary adenocarcinomas are a heterogenous group of tumours with poor prognosis that has
not improved considerably the last decades. Tumour morphology, i.e. intestinal type (I-type) or pancreatobiliary
type (PB-type), has been demonstrated to be a more relevant prognostic factor than anatomical origin. Tumour
infiltrating immune cells are vital in shaping the natural progress of cancer. The aim of this thesis was to
characterise the landscape of immune cells and common mutations in the entire spectrum of periampullary
adenocarcinoma, with particular reference to morphology and clinical outcome.
Methods: All studies are based on tumours from a retrospective, consecutive cohort of 175 patients with
periampullary adenocarcinoma who underwent surgical resection in Skåne University Hospital between 2001 and
2011. The infiltration of several immune cell populations was analysed by single-marker immunohistochemistry
(Paper I-II) and multiplexed immunofluorescence (Paper IV and V) on tissue microarrays. Targeted next
generation DNA seqencing (Paper III) was applied to analyse mutations in 70 common cancer-associated genes
in tumours from 102 cases.

Results: Paper I shows that high infiltration of natural killer NK/NKT cells was associated with prolonged overall
survival (OS), particularly in patients with I-type tumours. In patients with PB-type tumours, high infiltration of
NK/NKT cell infiltration was only prognostic in cases who did not receive adjuvant chemotherapy, and, notably,
there was a significant negative interaction between adjuvant chemotherapy and NK/NKT cell density in relation to
OS. Paper II shows that high infiltration of immature dendritic cells was an independent factor of poor prognosis in
patients with PB-type tymours. High infiltration of CD68+ and CD163+ macrophages was assocated with reduced
OS in the entire cohort, whereas high infiltration of MARCO+ macrophages was a negative prognostic factor only
in patients who recived adjuvant chemotherapy, but there was no signficant treatment interaction. Paper III
demonstrated that APC and ERBB3 mutations were more common in I-type tumours while CDKN2A mutations
were more common in PB-type tumours. KRAS mutations were associated with reduced OS in patients with I-type
tumours. in patients with PB-type tumours, SMARCA4 mutation was a negative prognostic factor in patients not
receiving adjuvant chemotherapy and there was a positive interaction between high expression of BRG1, the
protein encoded by SMARCA4, and adjuvant chemotherapy in relation to OS. Paper IV demonstrates that the
prognostic impact of different lymphocyte subsets, including signatures thereof, differ by morphology, and that
high levels of CD8+ T cells interacting with cancer cells and CD4+ T cells, respectively, were associated with a
prolonged OS. Moreover, immune cell density differed by the mutational status of several genes. Paper V provides
further validation of the beneficial prognostic impact of NK/NKT cells, also in the context of interaction with
macrophages. In addition, several prognostic innate immune cell subsets and signatures were defined, that
differed by tissue compartment and morphology.

Conclusions: The prognostic and potential predictive impact of tumour infiltrating immune cells and common
mutations in periampullary adenocarcinoma differs by morphology, thus highlighting the importance of taking
morphology into account in the quest for complementary biomarkers. Moreover, the prognostic value of tumour
infiltrating immune cells can be futher refined by analyses of their spatial and compartmental distribution.