Distinct Genomic Features Between Osteosarcomas Firstly Metastasing to Bone and to Lung

doi:10.21203/rs.3.rs-1354631/v1

PDF

Research Article

Distinct Genomic Features Between Osteosarcomas Firstly Metastasing to Bone and to Lung

https://doi.org/10.21203/rs.3.rs-1354631/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted 15 Feb, 2022

You are reading this latest preprint version

Background: Osteosarcoma (OS) initially metastasing to bone only shows distinct biological features compared to osteosarcoma that firstly metastasizes to the lung, which suggests us underlying different genomic pathogenetic mechanism. However related-study is lacking.

Methods: We analyzed whole-exome sequencing (WES) data for 38 metastatic OS patients with different relapse patterns. Multiplex Immunohistochemistry was also applied to investigate the immune- microenvironment. We tried to redefine disease subclassifications for OS based on genetic alterations to guide treatment strategies, and we correlated these genetic profiles with clinical treatment courses and prognosis to elucidate potential evolving cladograms.

Results: We investigated whole-exome sequencing of 12/38 patients with high-grade OS (31.6%) with initial bone metastasis (group A) and 26/38 (68.4%) with initial pulmonary metastasis (group B), of whom 15/38 (39.5%) had paired samples of primary lesions and metastatic lesions for these two cohorts with different relapse patterns. We found that osteosarcoma in group A mainly carries single-nucleotide variations while those in group B mainly exhibits structural variants. Additionally, osteosarcoma in group A displayed better immunogenicity, including a higher tumor mutation burden and neoantigen load and more tertiary lymphoid structures in tumor microenvironment.

Conclusions: Osteosarcoma with different relapse patterns has distinct genomic landscapes. This study suggests that osteosarcoma with mainly single-nucleotide variations other than structural variants might exhibit biological behavior predisposing toward bone metastases; immunotherapy is a promising therapeutic option for this group of patients.

Trial registration

This trial was registered in the Medical Ethics Committee of Peking University People’s Hospital on June 25^th, 2019 (Institutional Review Board, IRB, number of 2019PHB139-01). The trial was registered in Clinicaltrials.gov with identifier no. NCT03997747.

genomic sequencing

osteosarcoma

bone metastasis

single-nucleotide variation

structural variations

High-grade osteosarcoma (OS) is a bone-forming tumor characterized by the presence of an extracellular osteoid matrix produced by cancer cells [1]. Despite several decades of clinical trials, OS is still treated with pre- and postoperative chemotherapy as well as surgical resection of the tumor as definitive local treatment [2, 3]. The overall survival and event-free survival for patients with nonmetastatic disease at 5 years is approximately 71% and 54%, respectively [4, 5]. Cases of relapse with lung or bone metastasis fare less well, with only approximately 10% to 40% of patients alive at 5 years [4-6]. Treatment of relapse consists of radiation or surgery alone, with or without additional individualized systemic treatment [7]. Cases of multiple relapses have a poorer prognosis due to frequent second or subsequent recurrence [8, 9].

According to the series of Cooperative Osteosarcoma Study Group (COSS) trials, metastasis most commonly develops in the lungs (81.4%) and then in bones (7.8%) but rarely in lymph nodes, with a median time from diagnosis to relapse of 1.6 years (range, 0.1 to 14.3 years) [6, 10]. There is a small subset of OS patients who first experience relapse with distal skeletal metastasis alone, which has been assumed to be different with regard to clinical and biological features compared to those who experience first relapse involving the lung [10-12]. It is noteworthy that this population is also different from those who present with synchronous regional bone metastases (skip metastases) with an outlook extremely poor [13], as well as those with local recurrences involving disputable suspicion of former surgical margins [14]. According to Aung et al. [15] and Bacci et al. [10], the prognosis of patients who relapse with bone metastasis alone is worse than that of patients who relapse with lung metastasis. However, those with single, resectable and late-relapsed skeletal metastasis (usually ≥ 2 years) have almost the same or even better outcomes as patients with lung metastasis [10, 15]. Unfortunately, there is almost no basic research on exploration of the genetic occurrence and development of this group of OS due to the rarity of the disease.

Next-generation sequencing (NGS) has provided much insight into OS, which is characterized by significant structural variations (SVs), including copy number alteration (CNA), chromothripsis, kataegis and loss of heterozygosity (LOH), with few point mutations [16, 17]. This genomic landscape suggests that copy number (CN)-amplified genes may be critical drivers of OS progression and maintenance [17]. To date, relatively few studies have systematically evaluated the evolution of OS using matched samples [18] not to mention the population with initially only bone metastasis. This gap has motivated us to explore tumor biological characteristics and predictive biomarkers for personalized therapy for this unique subset.

Therefore, in this study, we analyzed whole-exome sequencing (WES) data for 38 metastatic OS patients, of whom 12 initially experienced relapse with skeletal metastasis (defined as group A). Eight of the 12 patients had paired samples (primary lesions and bone metastatic lesions) with which we were able to identify and compare differences between this cohort and 26 contemporary patient samples with lung metastasis for the first time (defined as group B), 7 of whom also had paired samples (primary lesions and pulmonary metastatic lesions). We also sought to redefine disease subclassifications for OS based on genetic alterations to guide treatment strategies, and we correlated these genetic profiles with clinical treatment courses and prognosis to elucidate potential evolving cladograms.

Study design

This study was performed after approval from the institutional review board of Peking University People’s Hospital(PKUPH). From May 1^st to October 1^st 2021, continuous patients who met the following criteria were included:1) histologically confirmed high-grade OS; 2) initial treatment in musculoskeletal tumor center of PKUPH; 3) operable metastatic lesions which later received resection in our center with samples available for WES. We retrospectively collected some of the matched primary samples in the pathology department of PKUPH for analysis. For samples collected after May 2021, fresh samples of pulmonary or osseous metastatic lesions as well as some paired fresh primary samples were also collected separately, which then performed both WES and RNA sequencing (RNA-seq). We excluded those patients without complete clinical information, those with unqualified specimens, those initially metastatic to other sites other than bone or lung, as well as those lost to follow-up. Finally, 38 OS patients with 58 samples were included in the analysis.

For bioinformatics analysis, we tried to (1) identify genomic aberrations between patients who firstly relapsed in skeletal metastasis (group A) and patients who relapsed with lung metastasis for the first time (group B), including all kinds of genomic alterations, tumor mutation burden (TMB), microsatellite instability (MSI), the number of neoantigens, human leukocyte antigen-I evolutionary divergence (HED) status and so on; (2) detect tumor heterogeneity between primary and their matched metastatic lesions for OS, especially in the type of genomic mutations, TMB, MSI, the number of neoantigens; (3) investigate the tumor evolution models and patients’ prognosis in different groups, including the overall survival and progression-free survival; (4) comprehensively investigate the tumor microenvironments (TME) spatial heterogeneity in selected cases, and additionally, to determine their relevance to tertiary lymphoid structures (TLS) using a multiplex immunohistochemistry (m-IHC) panel consisting of designed surface and intracellular markers (CD68, CD163, CD206, IRF8, CD3, CD8, PD-L1, CD20, CD56, B7-H3, CD47, CD4).

Tumor specimen processing and genomic DNA sequencing

Formalin-fixed, paraffin-embedded (FFPE) samples were used to produce 5-mm-thick slides. The pathologist evaluated the hematoxylin and eosin (HE) slides of each specimen and ensured that the tumor content was more than 20%. DNA from FFPE tumor tissues and matched blood was obtained by using QIAamp DNA FFPE Tissue Kit and QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany), respectively. Agarose gel electrophoresis was used to examine the quality and integrity of the DNA.

Genomic DNA was prepared for paired-end library construction using a QIAamp DNA FFPE Tissue Kit and QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany). The libraries were captured by using the NGS-based WES of OrigiMed (Shanghai, China). Targeted genomic regions were sequenced with an Illumina NovaSeq 6000.

Genomic alteration detection

Genomic alterations were identified as follows. Single nucleotide variations (SNVs) were identified by MuTect (v1.7). Indels were identified by using PINDEL (V0.2.5). The functional impact of genomic alterations was annotated by SnpEff3.0. CNV regions were identified by Control-FREEC (v9.7) with the following parameters: window = 50000 and step = 10000. Gene fusions were detected through an in-house-developed pipeline. Gene rearrangements were assessed by Integrative Genomics Viewer (IGV).

The CNV region was further analyzed by the Genomic Identification of Significant Targets in Cancer (GISTIC2.0) algorithm [19]. GISTIC2.0 assigned a G-score to each aberration, which was defined based on the amplitude and frequency of its occurrence across samples. Then, the q-value of aberrant regions was calculated, and the region with q-value < 0.25 was significant. –Log10 (q-value) ≥ 0.6 and ≤ –0.6 were assigned as the thresholds for amplification and deletion [20].

Tumor mutation burden

The TMB, defined as the total number of somatic/acquired mutations per coding area of a tumor genome (Mut/Mb), has emerged as a potential predictive biomarker of response to immune checkpoint inhibitors. It was calculated by counting the coding somatic mutations, including SNVs and indels, per megabase of the sequence examined in each patient.

Microsatellite instability

MSI is related to the functional loss of DNA mismatch repair deficiency (dMMR), manifesting as the length change of simple repeat sequences (microsatellites) resulting from base insertion or deletion. Per-sample microsatellite length distributions and MSI were called using MANTIS [21] with the recommended whole-exome settings and parameters (mrq = 25, mlq = 30, mlc = 30, mrr = 5).

Neoantigens

Neoantigens are tumor-specific antigens that arise due to somatic mutations in the DNA of tumor cells. We utilized NetMHCpan v4.0 to estimate neoantigens, in which the binding affinity of the mutant peptides was higher than that of the corresponding wild-type peptide, with a predicted binding strength less than 500 nmol/L.

Phylogenetic tree and tumor phylogenetics analysis

All SNVs were applied to shape phylogenetic trees by the lineage inference for cancer heterogeneity and evolution (LICHeE) method [22], which makes use of the somatic SNV patterns of samples and their variant allele frequencies (VAFs) as lineage markers to build the tree. “Root mutation” was defined as a mutation located on the trunk of the phylogenetic tree for each patient.

Mutational signature analysis

Somatic SNVs were divided into 96 trinucleotides (mutated base plus its sequence context) by 16 possible flanking nucleotide contexts. With the 30 published COSMIC mutation signatures [23] as a reference, we used deconstrucSigs (version 1.8.0) to obtain the signature probability matrix. The result was visualized in ggplot2 (version 3.3.2)

Subclonal evolution analysis

According to reference counts, variant counts, copy numbers, and VAF of each somatic mutation, we used PyClone to infer putative clonal clusters and estimate their cellular prevalence [24]. The results were applied for evolution analysis of subclones with CITUP (Conality Inference in Tumors Using Phylogeny) [25] and visualized by Timescpe (version 1.10).

Multiplex Immunohistochemistry

To investigate the immunologic landscape within GC, 6 full-face FFPE patient samples were randomly selected from group A (3 patients) and group B (3 patients) (Supplementary Table 1; see Methods section). H&E-stained tissue sections were respectively reviewed by two senior pathologists (S.D.H. and S.K.K.) to identify the tumor core (C, major body of the tumor mass), edge (E), margin (M), and nontumor normal (N), which we refer to regions of interest (ROIs). These represent tumor positions relative to surrounding tissue (Fig. 1a; see Methods). The serially sectioned tissue was stained with the m-IHC panel (Fig. 1b and c). A total of 1800 high-power fields (C: 52%, E: 13%, M: 21%, N: 14%) were imaged across all patient ROIs. A supervised image analysis system (QuPath) was used to obtain cell phenotyping data based on the pattern of marker expression (Fig. 4a-d).

m-IHC was achieved with a PANO 7-plex IHC kit (cat 10004100100; Panovue, Beijing, China). FFPE sections of samples were dewaxed and rehydrated. For the first and subsequent rounds, antigen retrieval was performed in EDTA (pH 9.0) buffer using a microwave (100–150 mW, 10 min). The slides were cooled to room temperature (RT), washed with TBST/0.1% Tween (3 times, 5 min) and incubated with H2O2 (3%) for 10 min. The slides were washed and blocked with blocking buffer for 10 min. Three panels were used to perform m-IHC. For panel 1, the primary antibody CD68 was incubated at RT for 30 min. The slides were washed, and the secondary antibody polymer anti-mouse/rabbit IgG was incubated at RT for 10 min. TSA dye (1:100) was applied for 10 min after washing. This was repeated three more times using antibodies against CD163, CD206 and interferon regulatory factor 8 (IRF8). A similar technological process was carried out for panel 2, which was composed of CD3, CD8, CD20, CD56 and PD-L1 (for observing TLSs), and for panel 3, which was composed of CD4, CD47 and B7-H3. Nuclei were stained with DAPI, and slides were mounted with medium.

Statistical analysis

Statistical analyses were performed using R software. The median and quartiles were used to describe demographic data and clinical outcomes.. T test was used to evaluate differences in normally distributed data, while Mann–Whitney U (Wilcoxon) test was used to assess differences in non-normally distributed data. The chi-square test was employed to evaluate differences in enumeration data. Overall survival and progression-free survival were estimated by Kaplan–Meier analysis. Differences were significant at p < 0.05.

Our cohort consisted of 12/38 patients with high-grade OS (31.6%) with initial bone metastasis (group A) and 26/38 (68.4%) with initial pulmonary metastasis (group B), of whom 15/38 (39.5%) had paired samples of primary lesions and metastatic lesions (Table 1 and Supplementary Table 1 and Fig. 1a-c). For our discovery set, WES was at a median depth of ∽591× (range, 268∽1239) using paired constitutional DNA from peripheral blood (Supplementary Table 2). Five of 15 (33.3%) paired tumor samples were derived from pretherapeutic biopsies. In other words, these samples presented the initial characteristics of their tumor status. We first assessed genomic complexity within this cohort along with clinical progression in these cases. These data were then partly integrated with transcriptome and immunostaining results to characterize the genomic landscape of our OS cohort.

Table 1. Clinical demographics of the 38 metastatic osteosarcomas.

Characteristic	Group A	Group B	Statistical method	p value
n (%)	12 (31.6)	26 (68.4)
Age at diagnosis (in years)			T test	0.060
Range	6-61	6-57
Median (Q1, Q3)	19.5 (13.25, 38.75)	14 (11, 17.5)
Gender, n (%)			Chi-square	1.000
Male	9 (23.7)	20 (52.6)
Female	3 (7.9)	6 (15.8)
Paired primary osteosarcoma, n (%)			Chi-square	0.033
Yes	8 (21.1)	7 (18.4)
No	4 (10.5)	19 (50.0)
Primary sample site, n (%)			Chi-square	0.280
Femur	8 (21.1)	16 (42.1)
Humerus	0 (0.0)	3 (7.9)
Tibia	2 (5.3)	6 (15.8)
Pelvis	1 (2.6)	1 (2.6)
Spine	1 (2.6)	0 (0.0)
Metastatic sample site, n (%)			Chi-square	0.000^*
Femur	2 (5.3)	3 (7.9)
Fibula	2 (5.3)	1 (2.6)
Tibia	0 (0.0)	2 (5.3)
Pelvis	4 (10.5)	0 (0.0)
Rib	1 (2.6)	0 (0.0)
Sacrum	1 (2.6)	0 (0.0)
Spine	2 (5.3)	1 (2.6)
Lung	0 (0.0)	19 (50.0)
Genomic type, n (%)			Chi-square	0.0180^a
SNV	10 (26.3)	11 (28.9)
SV or mixed	2 (5.3)	15 (39.5)
TMB			Wilcoxon	0.010^*
Range	1.1-32.9	0-14.2
Median (Q1, Q3)	4.85 (2.75, 11.98)	2.4 (1.38, 4.45)
MSI status, n (%)			Chi-square	0.316
MSS	11 (29.0)	26 (68.4)
MSI-H	1 (2.6)	0 (0.0)
Neoantigen			Wilcoxon	0.002^*
Range	57-2084	0-1539
Median (Q1, Q3)	743 (316.5, 1034.5)	128.5 (49, 200.5)
HED status, n (%)			Chi-square	1.000
Low	10 (26.3)	22 (57.9)
High	2 (5.3)	4 (10.5)
Adjuvant therapy, n (%)
First-line treatment	11 (29.0)	25 (65.8)	Chi-square	0.538
Second-line treatment	9 (23.7)	22 (57.9)	Chi-square	0.656
Third-line treatment	5 (13.2)	10 (26.3)	Chi-square	1.000
Fourth-line treatment	2 (5.3)	5 (13.2)	Chi-square	1.000
Vital status, n (%)			Chi-square	0.108
Alive	7 (18.4)	22 (57.9)
Dead	5 (13.2)	4 (10.5)
PFS (mo)			T test	0.596
Range	0-48.23	2.23-34.43
Median (Q1, Q3)	12.4 (7.45, 21.22)	13.95 (8.00, 19.75)
OS or follow-up time (mo)			T test	0.856
Range	7-59	9-54
Median (Q1, Q3)	23.5 (15.25, 46.5)	27 (16.75, 38.25)

Abbreviations: SNV, single nucleotide variation; Amp, amplification; Fus, fusion; TMB, tumor mutation burden; MSI, microsatellite instability; MSS, microsatellite stability; MSI-H, microsatellite instability-high; PFS, progression free survival; Mo, month; OS, overall survival.

HED, human leukocyte antigen-I evolutionary divergence. HED score> 7.887, defined as HED high; HED score ≤ 7.887, defined as HED low.

[Insert Table 1 here]

Clinicopathologic characteristics and genomic landscape

The clinical demographics of the 38 high-grade OS patients are summarized in Table 1. The clinical, pathologic, and predominant genomic landscape characteristics of all OSs with DNA sequencing are shown in Fig. 1a. The median age at diagnosis was 19.5 years for group A (Q1, Q3, 13.3, 38.8) and 14.0 years for group B (Q1, Q3, 11.0, 17.5). Males comprised the majority of the studied population, with 75.0% and 77.0% in both groups, respectively. For most of our patients, the primary lesions were located in the femur: 66.7% in group A and 61.5% in group B. For resected and metastatic samples, 4/12 (33.3%) were located in the pelvis, 2/12 (16.7%) in the femur, fibula and spine separately and 1/12 (8.3%) in the rib and sacrum in group A. When those patients developed pulmonary metastasis, it was usually multiple and scattered, such that the possibility for metastasectomy was low. In group B, although 19/26 (73.1%) were pulmonary lesions, 3/26 (11.5%), 2/26 (7.7%), 1/26 (3.8%) were later-developed femoral, tibial, fibular and spinal lesions respectively. Regarding the last-collected samples, 91.6% in group A showed progression upon first-line chemotherapy and 75.0% upon second-line systemic therapy; for group B, 96.2% progressed on first-line therapy and 84.6% on second-line therapy (detailed information in Table 1 and Supplementary Table 1). Therefore, subsequent WES data analyses were performed on 58 OS samples from 38 patients. With a median follow-up time of 26.0 months (Q1, Q3, 16.8, 39.0), 7/12 (58.3%) in group A were alive with disease while 22/26 (84.6%) in group B were alive. We did not find any obvious significant difference in progression-free survival (p=0.33) or overall survival (p=0.06) (Supplementary Fig. 1).

We defined driver mutations in established cancer genes, which mainly include SNVs and SVs, to describe the genomic landscape for the whole population. It is interesting that our cohorts seemed to show diverse biological behaviors based on the main genomic manifestations. Our population contained more point mutations other than historical high-risk OS genetic sequencings published [17, 18, 26, 27]. Consequently, we adopted a novel strategy to re-subclassify our tumors, restricting our focus to high-confidence variants (see Methods). A large portion of the group A were of the SNV type. Group B showed similar manifestation as reported for high-risk OS, involving genomic amplifications as well as fusions or mixed type (which we called the SV type), with few recurrent point mutations (Fig. 1d).

Somatic alterations detected by our methods in the 38 metastatic patients are shown in Fig. 1a and listed in Supplementary Table 2. Among genes commonly mutated, TP53 and MYC were identified in 22/58 and 14/58 patients respectively (57.9% and 36.8%; Fig. 1a; Supplementary Table 2). However, for this cohort of OS patients, we did not specifically design our panel to identify TP53 intron 1 rearrangements, as has recently been reported in OS [17], and more TP53 mutation might be missing due to the WES approach. We also identified alterations in MUC16 (17 mutations in 11 patients, 25.9%), NCOR1 (2 mutations in 9 patients, 23.7%), RAD21 (2 mutations in 9 patients, 23.7%), PTK2 (1 mutations in 8 patients, 21.1%), CDK4 (1 mutations in 7 patients, 18.4%), GLI1(2 mutations in 7 patients, 18.4%) and IGF1R (1 mutations in 7 patients, 18.4%) (Supplementary Table 2). Rb1 loss was only observed in 13.8% (8/58) of samples. Nevertheless, we noticed that for WES, only 14% (4399/31364) of the genomic alterations are reported mutations, whereas 86% (26965/31364) are not reported in the literatures (Fig. 1b); it is unclear whether these alterations influence the occurrence and evolution of OS.

Approximately 89.7% (52/58) of samples had CNAs and structural rearrangements. With respect to the former (Fig. 1a and Supplementary Table 2), amplifications at NCOR1 (n=13/58 samples; 22.4%), RAD21 (n=12/58 samples; 20.7%), CDK4 (n=10/58 samples; 17.2%) and MYC (n=19/58 samples; 32.8%) were the most frequent CNAs. Furthermore, PIK3CG mutation was mutually exclusive of RB1 alteration; overall, this pathway was altered in approximately 67.2% (39/58) of OS samples. Other less common regions of recurrent amplification are provided in Fig. 1a and Supplementary Table 2. We did not observe MDM2 amplification, as expected [26].

The burden of coding indels and substitutions across the 58 tumor samples varied from 2 to 1482 mutations per tumor (median 84; Supplementary Table 2). The highest coding mutation burden (32.9 mut/Mb) was observed in patient WJY, who initially presented with bone metastasis with a time interval from diagnosis to metastasis longer than 24 months and whose tumor exhibited MSI.

Diverse genomic characterization between osteosarcoma initially metastatic to bones and to lungs

We showed the mutation profiles between group A and group B in Fig. 2a. Surprisingly, group A mainly exhibited gene substitutions and short indels (green mark); group B mainly presented with gene amplifications, deletions and rearrangements (red mark). To describe this phenomenon more precisely, we reclassified our population into two genomic types: SNVs, accounting for more than 40% of all genomic alterations, which included gene substitutions and short indels; SVs, accounting for more than 40% of all genomic alterations, which included amplifications, homozygous deletions and breakpoints that disrupt genes or generate gene fusions. For those with both SNVs and SVs of more than 40%, we classified them as the mixed type. In fact, a large portion of our population had the mixed type (17/38, 44.7%) (Fig. 1c). Thus, to highlight patients with distinct genomic manifestations from group A, we further integrated the mixed type into the SV type. We compared all these patients’ genomic changes (see Table 2). Based on chi-square analysis, we observed obverse statistical disequilibrium for the differences between two groups (p=0.018), which suggests that the biological behavior from the perspective of genomic manifestations was distinct for these two groups of population. It should be noted that we further compared the TMB, neoantigen, MSI and HED status between the two groups. However, due to small sample sizes, we only observed diverse differences in TMB (p=0.01) and neoantigen status (p=0.0028) respectively (Fig. 2c and d).

Table 2. The difference of osteosarcoma patients’ genomic alterations between group A and group B.

	Group A	Group B	Statistical method	p value
n (%)	12 (31.6)	26 (68.4)
Genomic type, n (%)			Chi-square	0.0180^a
SNV	10 (26.3)	11 (28.9)
SV or mixed	2 (5.3)	15 (39.5)
TMB			Wilcoxon	0.0100^a
Range	1.1-32.9	0-14.2
Median (Q1, Q3)	4.85 (2.75, 11.98)	2.4 (1.38, 4.45)
Neoantigen			Wilcoxon	0.0028^a
Range	57-2084	0-1539
Median (Q1, Q3)	743 (316.5, 1034.5)	128.5 (49, 200.5)
MSI, n (%)			Chi-square	0.3160
MSS	11 (29.0)	26 (68.4)
MSI-H	1 (2.6)	0 (0.0)
HED-status, n (%)			Chi-square	1.0000
Low	10 (26.3)	22 (57.9)
High	2 (5.3)	4 (10.5)

Abbreviations: SNV, single nucleotide variation; SV, structural variants; Amp, amplification; Fus, fusion; TMB, tumor mutation burden; MSI, microsatellite instability; MSS, microsatellite stability; MSI-H, microsatellite instability-high.

HED human leukocyte antigen-I evolutionary divergence. HED score> 7.887, defined as HED high; HED score ≤ 7.887, defined as HED low.

The copy number variation region was further analyzed, and we found that CNV amplifications and deletions were more frequent in group B than in group A (Fig. 2b). The most common amplified regions involved 17p11.2, 5p15.2, 9q22.32, 1q21.2, 2q11.1, 8q24.21, 19q12, 9p24.1, 2q11.2, and 15q26.3, including the common driver genes MYC, CCNE1, JAK2, and IGF1R. In addition, 16p13.3, 22q13.33, 17p13.1, 17q25.3, 10q26.3, 19p13.11, 16q24.3, 8p21.3, 9q33.3, and 11p15.5 were the most common regions with copy number loss. The regions with increased copy number only included the driver genes of PTK2, JAK2, and CDK4 in group A.

The mutation burden of OS was similar reported as pancancer analysis [28]. However, the median TMB was 4.85 (Q1, Q3, 2.75, 11.98) and 2.4 (Q1, Q3, 1.38, 4.45) for group A and group B, respectively (Fig. 2c). Tumor neoantigen status can reflect immunogenicity, though the emergence of neoantigens was less than that in melanoma, and the median amount of neoantigen in group A was almost six times more than that in group B, at 743 (316.5, 1034.5) versus 128.5 (49, 200.5) (p=0.0028) (Fig. 2d). We compared all mutations in cell signaling pathway between the two groups and found differences, with most of the genes participating in the DDR pathway (p=0.03), and Notch pathway (p=0.03), with no difference in the cell cycle, TP53, IGF-beta, MYC, HIPPO, RTK or RAS pathways (Fig. 2e).

We analyzed the gene signature profiles of all 38 patients based on the literature for renal cell carcinoma or non-small cell lung cancer. However, mutation spectra revealed significant differences for signature 1 between the groups, mainly referring to age (Supplementary Fig. 2). Thus, we further compared the age distribution between these groups of patients with the Mann–Whitney U test but did not observe any significant difference, with a p value of 0.06 (Supplementary Fig. 3).

For SNVs, 65.8% (25/38) were nonsynonymous mutations, which might partly correlate with the higher expression of neoantigens in group A (Fig. 1b). In four of these SNV-type cases, the paired metastatic lesions were of the SV type. The number of SNVs called with high confidence ranged from 7 to 1482 (median, 150) per exome. Of these, a median of 30 SNV changes (range 4–272) within protein-coding regions potentially result in a functional product. For SV, 44.7% (17/38) were mixed mutations, including amplifications, fusions and long indels; only 1 sample carried a TP53 gene rearrangement, and 14 other rearrangements involved TP53, RB1, NTRK3, LRP1B and other genes in 9 samples. Regarding somatic CNAs, 248 amplification events in 22 samples with a median copy number of 7 (range 3~77) were detected. There were 20 events in 10 samples of gene deletion, including 15 gene deletions and 5 partial deletions in a genomic region of 7236~196391 bp.

High conservation of reported genetic sequencing over time in most osteosarcomas

For the 15 patients with paired primary and metastatic samples, we performed tumor phylogenetics to identify the basis of metastasis. Of the 15 patients, 8/15 (53.3%) were in group A and 7/15 (46.7%) in group B; their genomic manifestations of non-silent mutations are depicted in Fig. 1c. Other than prominent mutation landscape diversity between group A and group B (Fig. 3a), we did not observe any difference between primary and metastatic specimens for genomic landscapes (Fig. 1c), TMB (Supplementary Fig. 6a), neoantigens (Supplementary Fig. 6b) or profiling pathways (Supplementary Fig. 6c). We observed high conservation of reported genetic sequencing over time in both group A and group B. However, it was difficult to precisely define whether these cases involved linear or parallel progression models [29] (Supplementary Figs. 4 and 5). Most of our cases, regardless of group A or group B, were more similar to tumor self-seeding model, in which the primary tumor continually emits cells to a metastasis that was in fact seeded at early stages of tumorigenesis, masking evidence of parallel progression [29]. Phylogenetic trees inferred from genomic data may represent the evolutionary life history of individual tumors, as discussed in the literature [29]. In the present study, the phylogenetic relationship between the regional biopsies of a given sample was inferred from mutation data using LICHeE (Lineage Inference for Cancer Heterogeneity and Evolution) [30], PyClone [24] and CITUP (Conality Inference in Tumors Using Phylogeny) [25]. We employed two methods to describe these phylogenetics to identify the origin of metastasis (Supplementary Figs. 4 and 5), which exhibited multi-sample lineage trees based on the VAF of somatic SNVs and the subclonal evolution model (mutation- and copy number-based) respectively. Overall, 3/8 (37.5%) patients in group A had a more linear evolutionary pattern, whereas 7/7 (100%) patients in group B had a more parallel evolutionary pattern (Supplementary Figs. 4 and 5). We utilized the root/branch ratio (dN/dS ratio) for calculation [31] and found that groups A and B were not statistical different (p=0.46).

Treatment-induced amplifications

In this subset of the population, 3 patients (WS, HXD and WJY) in group A had treatment-naive primary and paired metastatic samples. For these three patients, two of their metastatic samples (WS and HXD) were resected after first-line chemotherapy according to our protocols [32]. The other patient, who was also the only patient (WJY) with a high TMB >30 mut/Mb, did not receive any systemic treatment other than resection during the whole disease process. In addition, the primary samples of two patients in group B (LZZ and SQZP) were treatment naive and derived from open incision biopsy before chemotherapy. However, these two cases both progressed so rapidly during treatment that the metastatic lesions had both progressed on first-line chemotherapy at the time of resection. Based on their genomic landscapes (Fig. 1c and Supplementary Figs. 4 and 5), we noticed that more genomic amplifications appeared after systemic treatment. For WS, bone metastatic lesions newly presented with 4 gene amplifications (ADGRA2, FGFR1, NSD3, POLB); for HXD, samples after treatment newly presented with PTK2 amplification, which did not occur in WJY, who did receive any systemic treatment for she was older than 40 years with relatively indolent disease process. For LZZ and SQZP, genomic amplifications were the major genomic alterations. Samples after systemic treatment showed a tendency toward a decline in SNVs. For the other paired samples, the primary lesions were resected after neoadjuvant chemotherapy; thus, we could not determine the influence of chemotherapy. It is suspected that systemic treatment, especially chemotherapy for OS, might contribute to genomic amplifications, which further might be the epitome of chromothripsis during tumor evolution.

We also compared their phylogenetic trees as well as clonal prevalence before and after chemotherapy (Supplementary Figs. 4 and 5) and did not find any relationship between evolutionary patterns and prognosis.

A distinctive immune ecosystem in osteosarcoma with initially bone metastasis based on multiplex immunohistochemistry

TLSs (tertiary lymphoid structures) are an important biomarker for identifying selected soft tissue sarcomas that might show a response to ICIs. On the basis of our patients’ genomic profiles, we next assessed tumor samples histologically to gain insight into the density and distribution of B cells as well as their relationship to other checkpoint molecules. Surprisingly, according to m-IHC of 3 randomly selected patient samples from group A, one (HXD) showed TLSs at the edges and margins of the tumor (shown in Fig. 3a2); 3 from group B lacked TLSs or CD3+CD8+ T cells (Fig. 3b2). It is also interesting that all patients in the vacant district of PD-L1 (programmed cell death ligand 1) expression usually showed high intensity and strongly positive expression of B7-H3 (Fig. 3a3, b3 and c1, p=0.0043). However, the patient (HXD) with multiple TLS districts of an immunogenic state exhibited poor staining of B7-H3 (Fig. 3a3), the mechanism of which should be further investigated. Notably, architectural analysis showed that CD20+ B cells were localized in TLSs and colocalized with CD4+, CD8+ and CD56+ T cells. Findings between gene expression profilings and immunohistochemistry analysis were complementary and showed modest correlation, which suggested that group A patients seemed to be in a more immunogenic state than group B. We later tried a single anti-PD-1 antibody (camrelizumab, Jiangsu HengRui Medicine Co., 200 mg ivgtt. Q2W) for another patient in our clinical practice in a multiple bone metastatic state but without any pulmonary metastasis after 4 lines of systemic treatment. Surprisingly, after just 4 infusions, a PET/CT scan showed an obvious a partial metabolic response (PMR) by PET Response Criteria in Solid Tumors (PERCIST) [33] (detailed case report in Supplementary Data 6 and Supplementary Fig. 7). Longer follow-up is needed to confirm the duration of the response and overall survival.

Tumor-associated macrophages (TAMs), one of the most abundant immune components in OS, are difficult to characterize due to their heterogeneity. Although multiple approaches have been used, the spatial distribution of TAMs in situ remains unclear. Thus, to examine the TAM distribution within the microenvironment, we analyzed the spatial density of the populations characterized above within different ROIs. The individual TAMs for all patients were then plotted based on the intensity of CD68, CD163, and CD206 (Fig. 4d and e), providing evidence of a spectrum of macrophage populations. The average intensity of each marker was determined on a per-patient basis, confirming that these populations are represented within each individual patient sample and ROI. An increase in overall macrophage density was observed within tumor regions (edge and margin, but not core, detailed in Supplementary Table 3) compared with adjacent normal tissue (Supplementary Fig. 8a, p=0.056).

The first site of metastasis in OS patients is usually the lung. Skeletal metastasis has rarely been described as having distinct biological and clinical behaviors and with diverse outcomes [10, 12, 15]. We conducted a comprehensive genomic and immune characterization of primary and paired-metastatic OS specimens to determine the molecular bases for OS first metastasizing to bone alone. Furthermore, to compare differences, we conducted WES for 26 contemporary patient samples firstly metastasizing to the lung in the same institution, of which 7 patients had paired samples of primary lesions as well as metastatic lesions. To our knowledge, this might be the largest sample with paired specimens to demonstrate the genomic landscape and potential evolving cladograms over time for this special group (referred to as group A in our study, Table 3). We observed diverse differences in genomic manifestations for these two groups of population. First and most importantly, enrichment of SNVs in bone metastatic patients and focal clustered SVs obviously occurred in those with pulmonary metastasis, suggesting that oncogenesis may be more driven by catastrophic chromothripsis events in OS patients with lung metastasis. Second, our set of 15 primary and metastatic matched OS samples showed high conservation of reported genetic sequencing profilings; thus, although OS shows significant genomic complexity with SVs and genomic instability, many of the genomic events are early events that remain stable over time. Third, high-intensity chemotherapy might induce a small subset of OSs to evolve toward CNAs, whereas fewer recurrent SNVs are found, resulting in difficulty for targeted therapy. Fourth, OS firstly metastasizing to bone might initially benefit from immunotherapy because it tends to have more non-reported point mutations and its TME manifests as a more inflamed immune state. We then tried ICIs alone bravely in one of these patients and obtained an immunological response within 12 weeks (Supplementary Fig. 7).

Table 3. List of recent progress made in next-generation sequencing (NGS) of osteosarcoma.

Year	First author	Organization	Patient number (N)	Sequencing methods^a	Main results
2013	Sharon A. Savage	National Cancer Institute of USA	941	multi-stage genome-wide association study (GWAS)	Two susceptibility loci warrant further exploration to uncover the biological mechanisms underlying susceptibility to osteosarcoma.
2014	Daniela Egas-Bejar	MD Anderson Cancer Center	20	182-gene next-generation exome sequencing (Foundation Medicine, Boston, MA)	The biology of aggressive and the metastatic phenotype osteosarcoma at the molecular level is similar to human fingerprints, in that no two tumors are identical.
2014	Xiang Chen	St. Jude Children’s Research Hospital	20	WGS	Recurrent Somatic Structural Variations Contribute to Tumorigenesis in Pediatric Osteosarcoma.
2014	Chaoyin Jiang	Shanghai Jiao Tong University Affiliated Sixth People’s Hospital	168	GWAS	GRM4 gene polymorphism is associated with susceptibility and prognosis of osteosarcoma in a Chinese Han population.
2014	Jennifer A. Perry	Dana–Farber/Boston Children’s Cancer	59	WES, WGS and RNA-seq	Complementary genomic approaches highlight the PI3K/mTOR pathway as a common vulnerability in osteosarcoma.
2015	Lisa Mirabello	National Cancer Institute of USA	935	GWAS	Variants in NFIB associated with metastasis in patients with osteosarcoma.
2015	Michal Kovac	University Hospital Basel, Switzerland	31	WES	Multiple oncogenic pathways drive chromosomal instability during osteosarcoma evolution and result in the acquisition of BRCA-like traits, which could be therapeutically exploited.
2016	M. Bousquet	Cancer Research Center of Toulouse, France	7	WES	Bone oncogenesis driven by TP53 or RB1 mutations occurs on a background of relative genetic stability and that the dedifferentiated OS subtype represents a clinico-pathological entity with distinct oncogenic mechanisms and thus requires different therapeutic management.
2017	Roelof Koster	National Cancer Institute of USA	632	GWAS	Genome-wide association study identifies the GLDC/IL33 locus associated with survival of osteosarcoma patients.
2017	Sam Behjati	Cancer Genome Project, UK	112	WGS, WES, RNA-seq	Recurrent mutation of IGF signalling genes and distinct patterns of genomic rearrangement in osteosarcoma. It may represent an age-independent mutational mechanism that contributes to the development of osteosarcoma in children and adults alike.
2018	Leanne C. Sayles	University of California	23	WGS	Using patient-derived tumor xenografts, we found a high degree of response for “genome-matched” therapies, demonstrating the utility of a targeted genome-informed approach.
2018	Huaiyuan Xu	the First Affiliated Hospital of Sun Yat-Sen University	13	WES	Loss of TP53 or RB1 is among the early events during OS tumorigenesis, while loss of PTEN is involved at the later stages associated with lung metastases. Finally, KEAP1 was identified as a novel biomarker for increased metastatic risk.
2019	Yoshiyuki Suehara	Memorial Sloan Kettering Cancer Center	66	MSK-IMPACT (Integrated Mutation Profiling of Actionable Cancer Targets) panel assay	Potentially clinically actionable alterations were found in approximately 21% of patients with osteosarcoma. At least 40% of patients have tumors harboring PDGFRA or VEGFA amplification. A new genomically based algorithm was proposed for directing patients with osteosarcoma to clinical trial options.
2020	Yan Zhou	Shanghai Jiao Tong University Affiliated Sixth People’s Hospital	11	scRNA-seq	Proinflammatory FABP4+ macrophages infiltration is noticed in lung metastatic osteosarcoma lesions. Lower osteoclasts infiltration is observed in chondroblastic, recurrent and lung metastatic osteosarcoma lesions compared to primary osteoblastic osteosarcoma lesions. Importantly, TIGIT blockade enhances the cytotoxicity effects of the primary CD3+ T cells with high proportion of TIGIT+ cells against osteosarcoma.
2020	Chia-Chin Wu	MD Anderson Cancer Center	48	WGS, RNA-seq and T-cell receptor sequencing	Neoantigen expression in OS was lacking and significantly associated with high levels of nonsense-mediated decay (NMD). Samples with low immune infiltrate had higher number of deleted genes while those with high immune infiltrate expressed higher levels of adaptive resistance pathways. PARP2 expression levels were significantly negatively associated with the immune infiltrate.
2021	Lu Xie and Zhenyu Cai (present study)	Peking University People’s Hospital	38 (15 patients with paired samples)	WES	Osteosarcoma initially metastasizing to bone mainly showed non-synonymous single-nucleotide variants, while those firstly relapsed with pulmonary metastasis mainly had structural variants.

^aIncluding whole-genome sequencing (WGS), whole-exome sequencing (WES), panel-based sequencing, total RNA/transcriptomic sequencing (RNA-seq), single cell RNA-seq (scRNA-seq), and proteomic sequencing

[Insert Table 3 here]

For the 12 patients in group A, only one female 61-year-old patient (WJY) had a time interval from diagnosis to bone metastasis longer than 24 months; according to the description of Bacci et al. [10] and Aung et al. [15], this patient should have a similar or even better prognosis than those with first metastasis to the lung. Interestingly, only this patient had TMB >30 mut/Mb with MSI-H (Supplementary Table 2), suggesting that immunotherapy might be effective. Nevertheless, this patient did not show high expression of PD-L1 or TLSs but only a high level of B7-H3; at the last follow-up, she was disease free after resection of the metastatic lesion. Thus, we have not been implemented any systemic therapy yet.

OS is characterized by complex genome instability and a high level of genetic heterogeneity [16]. OS is also characterized by specific alterations in tumor suppressors and oncogenes, including p53, Rb, CDK4, MDM2, and MYC [34, 35]. The majority of the resulting genetic alterations are associated with copy number changes and genome rearrangement [16, 17, 36, 37] (Table 3). Whether the observed SVs are truly targetable vulnerabilities remains to be fully explored [17], and clinically, it is almost certain that combination approaches need to be developed, as single agents are unlikely to lead to significant tumor regression. For many decades, treatment for OS has strongly focused on the development of cytotoxic and cytostatic drugs, frequently leading to disappointing results due to inter- or intratumoral heterogeneity [38]. The development of immunotherapy has revolutionized the status quo for inducing durable responses and improving OS [38]. However, in recent trials of immune checkpoint inhibitors (ICIs), only a limited number of OS patients have derived meaningful clinical benefits, without a statistical advantage for the whole population [39-43]. Selecting patients who may benefit from immunotherapies has become an urgent question that needs to be addressed. According to our study, those OS with initial bone metastasis alone might benefit from immunotherapy.

Several studies have shown that the nonsynonymous point mutation burden is one of the most reliable predictive biomarkers associated with responsiveness to ICIs in both melanoma and lung cancer [44-46]. Recently, Turajlic et al. [47] investigated whether the frameshift nature of indel mutations can create novel open reading frames and a large quantity of neoantigens, which might contribute to the immunogenic response. To further investigate the immune landscape and immunogenomic interplay for this group of populations, we used a m-IHC panel consisting of surface and intracellular designed markers, which indicated greater immunogenicity in our patient with bone metastasis. Thus, we hypothesize that for OS with initial skeletal metastasis, immunotherapy such as ICIs or combination therapy with ICIs might be a good option. We then further used anti-PD-1 therapy for one heavily treated advanced case with bone metastasis alone, and this patient achieved a partial response within 12 weeks, which needs to be confirmed with longer follow-up.

However, our methods involved WES for only selected cases including fresh specimens directly after surgical resection, as detected by RNA-seq. Wu et al. [27] found that unexpressed mutations tend to occur in genes that have low expression or with low VAFs. In addition, a highly mutated and rearranged OS genome may not generate sufficient neoantigens to elicit an immune response [27]. Thus, our WES only analysis might miss abundant genetic information for these patients. Biomarkers as well as appropriate immunotherapeutic strategies should be further verified in well-designed clinical trials.

OS first metastasizing to bone tended to show genetic profiling with nonsynonymous SNVs with higher TMB and more TLS expression, with obvious differences compared to OS first metastasizing to the lung. Although we sought to validate our assumption by larger sample sizes, the clinical information was not as meticulous and complete in The Cancer Genome Atlas (TCGA) and the Therapeutically Applicable Research To Generate Effective Treatments (TARGET) database with only scarce cases without statistical comparison (Supplementary Tables 4 and 5). Nevertheless, between our two groups of patients, we noticed that gene signature 1 had distinct expression, with a p value of 0.004, which indicated that the age distribution might confound the outcome. We further listed all these patients’ ages and compared their distribution between the two groups in an independent-sample t test and did not find a significant difference (p=0.06) (Supplementary Fig. 3). Several other studies [16, 48, 49] have revealed distinct molecular characteristics for pediatric and adult OSs with small sample sizes, however in actual fact this speculation has not been verified by direct evidence.

Several limitations in the current study should be acknowledged. First, most specimens were collected retrospectively from formalin-fixed paraffin-embedded slices; thus, we performed only WES without RNA-seq to investigate the process of tumor immune microenvironment evolution and to verify the WES results. Second, for primary lesions, only 33.3% (5/15) of the specimens were treatment-naïve; all the other specimens underwent combined chemotherapy before surgery following the PKUPH-02 protocol [50]. Thus, the influence of neoadjuvant chemotherapy on the gene expression could not be determined. However, these results might reflect the real status of OS encountered in clinical practice, and from our study, high conservation of reported genetic profiling was observed over time. This influence might not be as dominatingly controversial regarding outcome. Third, the sample size was relatively small; we attempted to test our hypothesis using data in public databases, but only scarce cases with clinical courses could be found. We could not collect a larger sample size with genetic materials. Therefore, we could not confirm the interpretation of our results. Fourth, we detected significant heterogeneity between the patients, and an independent comparison with more samples may help to further validate the results and to eliminate any possible bias due to heterogeneity. Finally, a set of markers was identified in this study, implying that immunotherapy might be effective for this special group, though we have not yet verified it in more patients in the form of clinical trials, which further need to be intelligently designed to confirm our hypothesis.

OS initially metastasizing to bone mainly showed more SNVs, whereas those first metastasizing to the lung mainly had SVs. In addition, OS initially metastasizing to bone showed better immunogenicity, including a higher TMB and higher neoantigen load, compared to those first metastasizing to the lung. This study suggests that OS patients with initial genetic landscapes of mainly SNVs should be paid more attention on monitoring bone metastases; immunotherapy might be a promising therapeutic option for this group of patients.

IRB, Institutional Review Board; OS, osteosarcoma; COSS, Cooperative Osteosarcoma Study Group; NGS, Next-generation sequencing; SV, structural alteration; CNA, copy number alteration; LOH, loss of heterozygosity; CN, copy number; WES, whole-exome sequencing; PKUPH, Peking University People’s Hospital; RNA-seq, RNA sequencing; TMB, tumor mutation burden; MSI, microsatellite instability; HED, human leukocyte antigen-I evolutionary divergence; TME, tumor microenvironments; TLS, tertiary lymphoid structures; m-IHC, multiplex immunohistochemistry; FFPE, Formalin-fixed, paraffin-embedded; HE, hematoxylin and eosin; IGV, Integrative Genomics Viewer; GISTIC, Genomic Identification of Significant Targets in Cancer; dMMR, DNA mismatch repair deficiency; LICHeE, lineage inference for cancer heterogeneity and evolution; VAFs, variant allele frequencies; CITUP, Conality Inference in Tumors Using Phylogeny; C, core; E, edge; M, margin; N, nontumor normal; RT, room temperature; IRF8, interferon regulatory factor 8; SNV, single nucleotide variation; TAM, Tumor-associated macrophage; ROIs, regions of interest; ICI, immune checkpoint inhibitor; TCGA, The Cancer Genome Atlas; TARGET, Therapeutically Applicable Research To Generate Effective Treatments;

Ethics approval and consent to participate

The study was in accordance with Declaration of Helsinki and International ethical guidelines of biomedical research. This trial was registered in the Medical Ethics Committee of Peking University People’s Hospital on June 25th, 2019 (Institutional Review Board, IRB, number of 2019PHB139-01). The trial was registered in Clinicaltrials.gov with identifier no. NCT03997747. All the patients included in this study had signed consents to participate in this study in Chinese.

Consent for publication

Consents for publication had been obtained from all the patients included in this study in Chinese.

Availability of data and materials

Data are available upon reasonable request. All the data belongs to Peking University People’s Hospital. URL: https://www. pkuph. cn/. All data generated or analyzed during this study are included in this article and its supplementary files. Under reasonable request, detailed information could be obtained by signing informed consent by e-mail: [email protected]

Competing interests

The authors declare that they have no competing interests.

Funding

Chinese National Natural Science Foundation (No. 81972509 and No. 82072970).

Authors' contributions

Conception and design: Xiaodong Tang and Wei Guo;

Development of methodology: Fanfei Meng, Xin Zhang, Hezhe Lu and Yunfan Wang;

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Lu Xie, Zhenyu Cai, Han Wang, Zhiye Du, Tingting Ren, Jie Xu and Xin Sun;

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Lu Xie, Zhenyu Cai, Fanfei Meng, Xin Zhang, Hezhe Lu, Jiuhui Xu and Jingbing Lou;

Pathological reviewing the slides of this study: Yunfan Wang and Tingting Ren;

Clinical evaluation of the study: Yuan Li, Lu Xie and Jie Xu;

Writing, review, and/or revision of the manuscript: Lu Xie, Zhenyu Cai, Hezhe Lu, Fanfei Meng, Xin Zhang, Xiaodong Tang and Wei Guo;

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Fanfei Meng, Xin Zhang, Xiaoxing Su and Yan Lei;

Study supervision: Xiaodong Tang and Wei Guo;

Other (participated in enrolling patients and looking at the data and tumor board review of the cases): Kun Luo, Jie Xu and Xin Sun;

Final approval of manuscript: Lu Xie, Zhenyu Cai, Hezhe Lu, Fanfei Meng, Xin Zhang, Kun Luo, Xiaoxing Su, Yan Lei, Jiuhui Xu, Jingbing Lou, Han Wang, Zhiye Du, Yunfan Wang, Yuan Li, Tingting Ren, Jie Xu, Xin Sun, Xiaodong Tang and Wei Guo.

Acknowledgements

The Berry Oncology Corporation had supported this study for multiplex immunohistochemistry of our specimens. We thank all the patients and their families for participating in this study. We also thank SKK and SDH for review and interpretation of all the of pathological slides and immunohistochemistry staining for this study.

Luetke A, Meyers PA, Lewis I, Juergens H. Osteosarcoma treatment-where do we stand? A state of the art review. Cancer Treat Rev. 2014;40:523-32.
Lagmay JP, Krailo MD, Dang H, Kim A, Hawkins DS, Beaty O, 3rd et al. Outcome of patients with recurrent osteosarcoma enrolled in seven phase II trials through children's cancer group, pediatric oncology group, and children's oncology group: learning from the past to move forward. J Clin Oncol. 2016;34:3031-8.
Marina NM, Smeland S, Bielack SS, Bernstein M, Jovic G, Krailo MD et al. Comparison of MAPIE versus MAP in patients with a poor response to preoperative chemotherapy for newly diagnosed high-grade osteosarcoma (EURAMOS-1): an open-label, international, randomised controlled trial. Lancet Oncol. 2016;17:1396-408.
Smeland S, Bielack SS, Whelan J, Bernstein M, Hogendoorn P, Krailo MD et al. Survival and prognosis with osteosarcoma: outcomes in more than 2000 patients in the EURAMOS-1 (European and American osteosarcoma study) cohort. Eur J Cancer. 2019;109:36-50.
Arshi A, Sharim J, Park DY, Park HY, Yazdanshenas H, Bernthal NM et al. Prognostic determinants and treatment outcomes analysis of osteosarcoma and ewing sarcoma of the spine. Spine J. 2017;17:645-55.
Bielack SS, Kempf-Bielack B, Delling G, Exner GU, Flege S, Helmke K et al. Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol. 2002;20:776-90.
Meazza C, Bastoni S, Scanagatta P. What is the best clinical approach to recurrent/refractory osteosarcoma? Expert Rev Anticancer Ther. 2020;20:415-28.
Tirtei E, Asaftei SD, Manicone R, Cesari M, Paioli A, Rocca M et al. Survival after second and subsequent recurrences in osteosarcoma: a retrospective multicenter analysis. Tumori. 2018;104:202-6.
Bielack SS, Kempf-Bielack B, Branscheid D, Carrle D, Friedel G, Helmke K et al. Second and subsequent recurrences of osteosarcoma: presentation, treatment, and outcomes of 249 consecutive cooperative osteosarcoma study group patients. J Clin Oncol. 2009;27:557-65.
Bacci G, Longhi A, Bertoni F, Briccoli A, Versari M, Pignotti E et al. Bone metastases in osteosarcoma patients treated with neoadjuvant or adjuvant chemotherapy: the rizzoli experience in 52 patients. Acta Orthop. 2006;77:938-43.
Pirker-Fruhauf UM, Friesenbichler J, Rabitsch K, Liegl-Atzwanger B, Bauernhofer T, Windhager R et al. Three metachronous osteosarcomas within 22 years without pulmonary metastases: a case report and review of the literature. Case Rep Orthop. 2013;2013:197287.
San-Julian M, Diaz-de-Rada P, Noain E, Sierrasesumaga L. Bone metastases from osteosarcoma. Int Orthop. 2003;27:117-20.
Kager L, Zoubek A, Kastner U, Kempf-Bielack B, Potratz J, Kotz R et al. Skip metastases in osteosarcoma: experience of the cooperative osteosarcoma study group. J Clin Oncol. 2006;24:1535-41.
Picci P, Sangiorgi L, Rougraff BT, Neff JR, Casadei R, Campanacci M. Relationship of chemotherapy-induced necrosis and surgical margins to local recurrence in osteosarcoma. J Clin Oncol. 1994;12:2699-705.
Aung L, Gorlick R, Healey JH, Shi W, Thaler HT, Shorter NA et al. Metachronous skeletal osteosarcoma in patients treated with adjuvant and neoadjuvant chemotherapy for nonmetastatic osteosarcoma. J Clin Oncol. 2003;21:342-8.
Chen X, Bahrami A, Pappo A, Easton J, Dalton J, Hedlund E et al. Recurrent somatic structural variations contribute to tumorigenesis in pediatric osteosarcoma. Cell Rep. 2014;7:104-12.
Sayles LC, Breese MR, Koehne AL, Leung SG, Lee AG, Liu HY et al. Genome-informed targeted therapy for osteosarcoma. Cancer Discov. 2019;9:46-63.
Negri GL, Grande BM, Delaidelli A, El-Naggar A, Cochrane D, Lau CC et al. Integrative genomic analysis of matched primary and metastatic pediatric osteosarcoma. J Pathol. 2019;249:319-31.
Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol. 2011;12:R41.
Brocca G, Ferraresso S, Zamboni C, Martinez-Merlo EM, Ferro S, Goldschmidt MH et al. Array comparative genomic hybridization analysis reveals significantly enriched pathways in canine oral melanoma. Front Oncol. 2019;9:1397.
Kautto EA, Bonneville R, Miya J, Yu L, Krook MA, Reeser JW et al. Performance evaluation for rapid detection of pan-cancer microsatellite instability with MANTIS. Oncotarget. 2017;8:7452-63.
Ricketts C, Popic V, Toosi H, Hajirasouliha I. Using LICHeE and BAMSE for reconstructing cancer phylogenetic trees. Curr Protoc Bioinformatics. 2018;62:e49.
Tate JG, Bamford S, Jubb HC, Sondka Z, Beare DM, Bindal N et al. COSMIC: the catalogue of somatic mutations in cancer. Nucleic Acids Res. 2019;47:D941-7.
Roth A, Khattra J, Yap D, Wan A, Laks E, Biele J et al. PyClone: statistical inference of clonal population structure in cancer. Nat Methods. 2014;11:396-8.
Malikic S, McPherson AW, Donmez N, Sahinalp CS. Clonality inference in multiple tumor samples using phylogeny. Bioinformatics. 2015;31:1349-56.
Kovac M, Blattmann C, Ribi S, Smida J, Mueller NS, Engert F et al. Exome sequencing of osteosarcoma reveals mutation signatures reminiscent of BRCA deficiency. Nat Commun. 2015;6:8940.
Wu CC, Beird HC, Andrew Livingston J, Advani S, Mitra A, Cao S et al. Immuno-genomic landscape of osteosarcoma. Nat Commun. 2020;11:1008.
Cortes-Ciriano I, Lee JJ, Xi R, Jain D, Jung YL, Yang L et al. Comprehensive analysis of chromothripsis in 2,658 human cancers using whole-genome sequencing. Nat Genet. 2020;52:331-41.
Naxerova K, Jain RK. Using tumour phylogenetics to identify the roots of metastasis in humans. Nat Rev Clin Oncol. 2015;12:258-72.
Popic V, Salari R, Hajirasouliha I, Kashef-Haghighi D, West RB, Batzoglou S. Fast and scalable inference of multi-sample cancer lineages. Genome Biol. 2015;16:91.
Williams MJ, Werner B, Heide T, Curtis C, Barnes CP, Sottoriva A et al. Quantification of subclonal selection in cancer from bulk sequencing data. Nat Genet. 2018;50:895-903.
Xie L, Xu J, Li X, Zhou Z, Zhuang H, Sun X et al. Complete remission of metastatic osteosarcoma using combined modality therapy: a retrospective analysis of unselected patients in China. BMC Cancer. 2021;21:337.
O JH, Lodge MA, Wahl RL. Practical PERCIST: a simplified guide to PET response criteria in solid tumors 1.0. Radiology. 2016;280:576-84.
Wei G, Lonardo F, Ueda T, Kim T, Huvos AG, Healey JH et al. CDK4 gene amplification in osteosarcoma: reciprocal relationship with INK4A gene alterations and mapping of 12q13 amplicons. Int J Cancer. 1999;80:199-204.
Miller CW, Aslo A, Won A, Tan M, Lampkin B, Koeffler HP. Alterations of the p53, Rb and MDM2 genes in osteosarcoma. J Cancer Res Clin Oncol. 1996;122:559-65.
Martin JW, Squire JA, Zielenska M. The genetics of osteosarcoma. Sarcoma. 2012;2012:627254.
Zhou Y, Yang D, Yang Q, Lv X, Huang W, Zhou Z et al. Single-cell RNA landscape of intratumoral heterogeneity and immunosuppressive microenvironment in advanced osteosarcoma. Nat Commun. 2020;11:6322.
Galon J, Bruni D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat Rev Drug Discov. 2019;18:197-218.
Boye K, Longhi A, Guren T, Lorenz S, Naess S, Pierini M et al. Pembrolizumab in advanced osteosarcoma: results of a single-arm, open-label, phase 2 trial. Cancer Immunol Immunother. 2021;70:2617-24.
Tawbi HA, Burgess M, Bolejack V, Van Tine BA, Schuetze SM, Hu J et al. Pembrolizumab in advanced soft-tissue sarcoma and bone sarcoma (SARC028): a multicentre, two-cohort, single-arm, open-label, phase 2 trial. Lancet Oncol. 2017;18:1493-501.
D'Angelo SP, Mahoney MR, Van Tine BA, Atkins J, Milhem MM, Jahagirdar BN et al. Nivolumab with or without ipilimumab treatment for metastatic sarcoma (alliance A091401): two open-label, non-comparative, randomised, phase 2 trials. Lancet Oncol. 2018;19:416-26.
Xie L, Xu J, Sun X, Guo W, Gu J, Liu K et al. Apatinib plus camrelizumab (anti-PD1 therapy, SHR-1210) for advanced osteosarcoma (APFAO) progressing after chemotherapy: a single-arm, open-label, phase 2 trial. J Immunother Cancer. 2020;8:e000798.
Le Cesne A, Marec-Berard P, Blay JY, Gaspar N, Bertucci F, Penel N et al. Programmed cell death 1 (PD-1) targeting in patients with advanced osteosarcomas: results from the PEMBROSARC study. Eur J Cancer. 2019;119:151-7.
Van den Eynden J, Jimenez-Sanchez A, Miller ML, Larsson E. Lack of detectable neoantigen depletion signals in the untreated cancer genome. Nat Genet. 2019;51:1741-8.
Desrichard A, Snyder A, Chan TA. Cancer neoantigens and applications for immunotherapy. Clin Cancer Res. 2016;22:807-12.
Schumacher TN, Schreiber RD. Neoantigens in cancer immunotherapy. Science. 2015;348:69-74.
Turajlic S, Litchfield K, Xu H, Rosenthal R, McGranahan N, Reading JL et al. Insertion-and-deletion-derived tumour-specific neoantigens and the immunogenic phenotype: a pan-cancer analysis. Lancet Oncol. 2017;18:1009-21.
Suehara Y, Alex D, Bowman A, Middha S, Zehir A, Chakravarty D et al. Clinical genomic sequencing of pediatric and adult osteosarcoma reveals distinct molecular subsets with potentially targetable alterations. Clin Cancer Res. 2019;25:6346-56.
Staley BS, Milko LV, Waltz M, Griesemer I, Mollison L, Grant TL et al. Evaluating the clinical utility of early exome sequencing in diverse pediatric outpatient populations in the North Carolina clinical genomic evaluation of next-generation exome sequencing (NCGENES) 2 study: a randomized controlled trial. Trials. 2021;22:395.
Xie L, Xu J, Sun X, Li X, Liu K, Liang X et al. Apatinib plus ifosfamide and etoposide for relapsed or refractory osteosarcoma: a retrospective study in two centres. Oncol Lett. 2021;22:552.

No competing interests reported.

PDF

Version 1

posted 15 Feb, 2022

You are reading this latest preprint version

Distinct Genomic Features Between Osteosarcomas Firstly Metastasing to Bone and to Lung

Status:

Version 1

Abstract

Figures

Background

Methods

Results

Discussion

Conclusions

List Of Abbreviations

Declarations

References

Competing Interests

Supplementary Files

Status:

Version 1