br unique example wherein CPD and PPM D were
unique example, wherein CPD and PPM1D were co-amplified with ERBB2 in single nuclei but did not co-amplify with each other.
The amplification of ESR1 (6q25) was observed in six tu-mors including a case with an incidental metaphase spread, where four copies of ESR1 were seen arranged in tandem near
a centromere-6 signal (Supplementary Fig. 1). CCNE1 (19q12) was amplified in three cases: one such case exhibited the HSR type and the other two exhibited the LA type. Ampli-fication of AURKA (20q13.2) was found in two cases: one
such case exhibited the CoPoly type and the other exhibited the LA type. EGFR (7p11.2) and CDH1 (16q22.1) were am-plified in HSR in a separate case.
3.8. Co-amplifications of non-syntenic genes (genes physically located on different chromosomes)
Co-amplification of non-syntenic genes in single nuclei was found in 35 tumors, as listed in Table 2. Dual-color FISH revealed co-localization of an amplicon on 8p11 (con-sisting of ZNF703, FGFR1, ADAM9, and/or IKBKB) and an amplicon on 11q13 (including CCND1 with or without C11ORF30) in single BODIPY 493/503 in 10 tumors. In six of these spec-imens (Cases 1–6), the two amplicons were thought to consti-tute a single amplification unit because signals for the genes representing both amplicons were superimposed, as shown in Fig. 5A and B. However, in the other four of these 10 tumors, the signals for the two amplicons were separately distributed in individual nuclei, as shown in Fig. 5C. In Cases 3 and 4, sig-nals for centromeres 8 and 11 were detected in the amplifica-tion unit (Supplementary Fig. 2). A possible amplification status for Case 3, as inferred from these dual-color FISH find-ings on touch smears, is depicted schematically in Fig. 5D. Unfortunately, touch smears were not obtained for Case 4, pre-cluding this sort of precise analysis.
The amplicon containing ERBB2 and its flanking genes was co-localized with the amplicon on 8p11 in eight tumors, and with the amplicon on 11q13 in seven tumors, as shown in Table 2. In these tumors (excluding Case 14), the two ampli-cons were observed separately in individual nuclei (Fig. 5E). In Case 14, dual-color FISH on the touch smear revealed not only representative signals for the two amplicons (as shown in Fig. 5F and G) but also clustered signals for centromeres 8 and 11 (as shown in Supplementary Fig. 3). Based on these findings, the amplification status of Case 14 is inferred to be as depicted schematically in Fig. 5H.
In Case 25, clustered signals for the amplicon on 17q were co-amplified with CDH1, a gene originally located at 16q22.1 (Supplementary Fig. 4). Other co-amplifications of non-syntenic genes were observed in single nuclei but appeared to correspond to different amplicons.
In the present study, each of the 22 genes examined were found to be amplified in at least one of the tumor samples, al-though the frequency of detection of each gene varied. Some of these genes are established driver oncogenes that have al-ready been tested as clinical targets of molecular therapies; others are candidate oncogenes that are being targeted by ther-apies that are the subject of ongoing clinical trials or are in de-velopment. The targeted therapy against amplified ERBB2 has been the most successful molecular therapy in treating breast cancer. Clinical studies using FGFR1-selective inhibitors also have been performed [10,16-18]. CCND1 is generally regarded as difficult to target directly by therapies; however, CCND1’s partner kinases, such as cyclin-dependent kinases 4 and 6, are considered better targets [10,19,20]. Therapies targeting IKBKB and ADAM9 by the use of specific micro-RNAs are under investigation and remain the subject of in vitro studies [21,22]. ZNF703, the product of one of the genes most fre-quently amplified on 18p11, also is considered a promising can-didate for molecular targeting . In addition, although TOP2A is not a molecular target, TOP2A amplification is a bio-marker that predicts chemo-sensitivity to anthracyclines . Multiple other genes on the various amplicons also may have significant roles in breast cancer and so represent potential tar-gets for therapy.
Recently, a Phase II study evaluating the efficacy of doviti-nib, a potent FGFR inhibitor, against breast cancer terminated early because of slow accrual of patients with amplifications of FGFR1, FGFR2, and FGFR3 . To avoid delays in the re-cruitment of patients with rare molecular markers, molecular screening using next-generation sequencing assays  or droplet digital PCR analyses using circulating tumor DNA
 have been proposed. However, compared to these methods, MLPA is a relatively cheap and easy-to-perform PCR assay that allows simultaneous detection of multiple gene copy-number aberrations from small amounts of fragmented DNA derived from formalin-fixed material.
We also observed the close association of centromere signals with signals for the amplicons; specifically, the amplicons on 17q12–21, 11q13, 8p11, and 8q24 were closely associated with the respective centromere signals in five, two, one, and one tu-