Do you always have to return a promisified object in order to make sure its execution is fulfilled?

Question

Let's say you promisify a MySQL connection object (in order to make raw MySQL queries):

var con = Promise.promisify(MySQL.newCon());

And then, you start a promise chain with some ORM method. Inside of the first then, you get a color object, returned from the Color model, but then, you want to start a MySQL transaction AND pass that color to the next then in the promise chain (but also ensure that the transaction starts):

Color
    .findOne() // ORM looks for a color within the model (table)
    .then(function (color) {
        // A transaction is started
        // Whatever this query returns is useless,
        // Since it starts a transaction.
        // I'd like to actually pass the color to the next `then`.
        return con.query('START TRANSACTION');
    })
    .then(function (color) {
        console.log(color);
    });

I'm thinking that returning con.query ensures that the transaction is started (and the promise chain continues), but is that the only way? I know I could do:

return new Promise(function (resolve, reject) {
    con.query('START TRANSACTION', function (err) {
        if (err) {
            return reject(new Error(err));
        }
        resolve(color); // Pass the color to the next then
    });
});

Or could also return an array and use spread instead of then:

return [color, con.query('START TRANSACTION')];

Are these the only ways of doing this? Ideally, I'd like something that is very similar to the first example I posted, because I think that returning a new Promise looks dirty, and returning an array is kind of confusing, since I don't care about what con.query returns at all.

Thanks.