If your input file is sorted in the right way, then each line can just be processed as it is read. Since your test data set was sorted in that way, I took advantage of that to simplify the code. If that's not a good assumption, then of course things get more complicated.

#!/usr/bin/perl
use strict;
use warnings;
$|=1;

my $line =<DATA>;
my ($a,$b,$c,$d,$cnt,$str,$end) = ($line =~ m/:(\d+)/g); 

while (my $line2 = <DATA>)
{
    my ($A,$B,$C,$D,$Cnt,$Str,$End) = ($line2 =~ m/:(\d+)/g);
    if ($a == $A and $d == $D)
    {
         $b ="multiple"  if ($b ne $B);
         $c ="multiple"  if ($c ne $C);
         $cnt+=$Cnt;
         $end=$End;
         next;
    }
    print "A:$a  B:$b  C:$c  D:$d CNT:$cnt STR:$str END:$end\n";
    ($a,$b,$c,$d,$cnt,$str,$end)=($A,$B,$C,$D,$Cnt,$Str,$End);
}

print "A:$a  B:$b  C:$c  D:$d CNT:$cnt STR:$str END:$end\n";   

=prints with your test dataset:
A:1  B:multiple  C:3  D:4 CNT:3 STR:1 END:4
A:2  B:2  C:3  D:5 CNT:6 STR:4 END:10
A:3  B:2  C:3  D:4 CNT:1 STR:11 END:12

with another dataset of:
A:1  B:2  C:3 D:4  CNT:1 STR:1 END:2
A:1  B:7  C:3 D:4  CNT:1 STR:2 END:3
A:1  B:2  C:3 D:8  CNT:1 STR:3 END:4
A:2  B:2  C:3 D:5  CNT:1 STR:4 END:5
A:2  B:2  C:3 D:5  CNT:5 STR:5 END:10
A:3  B:2  C:3 D:4  CNT:1 STR:11 END:12

that prints:
A:1  B:multiple  C:3  D:4 CNT:2 STR:1 END:3
A:1  B:2  C:3  D:8 CNT:1 STR:3 END:4
A:2  B:2  C:3  D:5 CNT:6 STR:4 END:10
A:3  B:2  C:3  D:4 CNT:1 STR:11 END:12

=cut

__DATA__
A:1  B:2  C:3 D:4  CNT:1 STR:1 END:2
A:1  B:7  C:3 D:4  CNT:1 STR:2 END:3
A:1  B:2  C:3 D:4  CNT:1 STR:3 END:4
A:2  B:2  C:3 D:5  CNT:1 STR:4 END:5
A:2  B:2  C:3 D:5  CNT:5 STR:5 END:10
A:3  B:2  C:3 D:4  CNT:1 STR:11 END:12
[download]