A commenter suggests
It's simply an artifact of Democratic success in 2012
This is not true. The Senate class of 2012 was already unbalanced as a result of the pro-Democrat, anti-Republican race of 2006 and the somewhat balanced race of 2000. Prior to 2012, the ratio was twenty-three Democrats to ten Republicans. After, twenty-five to eight (including Angus King and Bernie Sanders who ran as independents in 2012), a gain of two.
The 2010 class was a good year for Republicans with a net gain of six seats over 2004. The 2016 class took the balance back by two seats. Currently twenty-two Republicans to eleven Democrats.
The 2014 class also produced a major shift, as it followed the good Democrat class of 2008. Currently twenty-two Republicans to eleven Democrats.
As you can see, while the 2012 class is the most unbalanced, every current Senate class is unbalanced.
Rep Dem Net Rep
1994 19 14 + 5
1996 21 12 + 9
1998 16 18 - 2
2000 14 19 - 4
2002 21 12 + 9
2004 19 15 + 4
2006 9 24 -15
2008 13 20 - 7
2010 24 10 +14
2012 8 25 -17
2014 22 11 +11
2016 22 12 +10
Note that these are the results of only that election class. If special elections were held for other election classes, their results were not included. Dem results include independents who caucus with the Democrats. Rep is just Republicans.
There are some competing causes here. Republicans tend to do better in midterms than presidential elections. Both parties tend to do better when they have a presidential candidate winning. Both parties do better in midterms when the other party has the presidency. In the period from 1998 to 2004, most of the Senate results were balanced.
In 2006, there was an unpopular Republican president and it was a midterm election. Two out of three advantages to the Democrats from a balanced start. Democrats were down one seat and gained six (including two independents, Bernie Sanders and Joe Lieberman). In 2012, Democrats gained additional seats in a presidential year where their candidate won.
In each of 2010 and 2014, there was a Democratic president and a midterm. All advantages to the Republicans. They started up by two seats and added six in 2010. They started down by seven and gained nine in 2014.
Things could go either way in 2018. It is the most unbalanced Senate class, which would tend to favor the Republicans. And it's a midterm: mild benefit for Republicans. But there will also be a president who ran as a Republican, generally an advantage for Democrats. Ten Democrats up for election in 2018 are in states that Donald Trump won in 2016. Only Dean Heller of Nevada is a Republican from a state that Hillary Clinton won. Jeff Flake from Arizona and Ted Cruz from Texas are Republicans from states where Clinton did better than expected. Cruz may struggle with Republicans after 2016.
Prior to 2012, most prognosticators expected Republican gains since the Democrats had few opportunities. But Democrats made gains. Maybe they'll do so again. Maybe not. Best guess might be that Democrats will lose a few seats to Republicans but end up with a smaller lead. But it's possible that they might gain seats or lose enough to lose the lead in this class. Republicans need eight for a supermajority of sixty seats.
A three seat gain for the Democrats (Heller, Flake, and Cruz) is possible but unlikely. An eleven seat gain for Republicans is possible but still unlikely (e.g. states they won in 2016 plus Minnesota where they almost won). More likely results are a one seat gain for Democrats (e.g. Heller), a two seat gain for Republicans (e.g. lose Heller but gain the seats of Claire McCaskill of Missouri, Jon Tester of Montana, and Joe Donnelly of Indiana; three Democrats known for the weakness of their 2012 Republican opponents), or a nine seat gain (e.g. every state votes for Senator as it voted for president in 2016). And of course everything in between.